Based on equalization image processing and spatial crosstalk attenuator
By training an equalizer to generate a lookup table, the problem of spatial crosstalk in pyrosequencing systems was solved, improving the accuracy of base detection and DNA sequencing.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ILLUMINA INC
- Filing Date
- 2021-05-05
- Publication Date
- 2026-06-30
Smart Images
Figure CN122312431A_ABST
Abstract
Description
[0001] This application is a divisional application of the application filed on May 5, 2021, with application number 202180029821.6 and invention title "Image Processing and Spatial Crosstalk Attenuator Based on Equalization".
[0002] Priority application
[0003] This PCT patent application claims the benefit of the following patent applications: U.S. Provisional Patent Application No. 63 / 020,449, filed May 5, 2020, entitled “EQUALIZATION-BASED IMAGE PROCESSING AND SPATIAL CROSSTALK ATTENUATOR” (Attorney’s File No. ILLM 1032-1 / IP-1991-PRV), and U.S. Patent Application No. 17 / 308,035, filed May 4, 2021, entitled “EQUALIZATION-BASED IMAGE PROCESSING AND SPATIAL CROSSTALK ATTENUATOR” (Attorney’s File No. ILLM 1032-2 / IP-1991-US). The priority applications are incorporated herein by reference for all purposes. Technical Field
[0004] The techniques disclosed in this invention relate to apparatus and methods for automatically analyzing images or recognizing patterns. This includes systems for transforming images for the purposes of: (a) enhancing the visual quality of an image prior to recognition; (b) locating and recording an image relative to a sensor or stored prototype, or reducing the amount of image data by discarding irrelevant data; and (c) measuring the salient features of an image. Specifically, the techniques disclosed in this invention relate to removing spatial crosstalk from sensor pixels using equalization-based image processing techniques.
[0005] Literature merged
[0006] The following references are incorporated by way of citation, i.e., as shown in their entirety in this document, for all purposes:
[0007] U.S. non-provisional patent application No. 15 / 936,365, filed on March 26, 2018, entitled “DETECTION APPARATUS HAVING AMICROFLUOROMETER, A FLUIDIC SYSTEM, AND A FLOW CELL LATCH CLAMP MODULE”;
[0008] U.S. non-provisional patent application No. 16 / 567,224, entitled “FLOW CELLS AND METHODS RELATED TO SAME”, filed on September 11, 2019;
[0009] U.S. non-provisional patent application No. 16 / 439,635, entitled “DEVICE FOR LUMINESCENT IMAGING”, filed on June 12, 2019;
[0010] U.S. non-provisional patent application No. 15 / 594,413, filed on May 12, 2017, entitled “INTEGRATED OPTOELECTRONIC READ HEAD AND FLUIDIC CARTRIDGE USEFUL FOR NUCLEIC ACID SEQUENCING”;
[0011] U.S. non-provisional patent application No. 16 / 351,193, filed on March 12, 2019, entitled “ILLUMINATION FOR FLUORESCENCE IMAGING USINGOBJECTIVE LENS”;
[0012] U.S. non-provisional patent application No. 12 / 638,770, filed on December 15, 2009, entitled “DYNAMIC AUTOFOCUS METHOD AND SYSTEM FORASSAY IMAGER”;
[0013] U.S. non-provisional patent application No. 13 / 783,043, filed on March 1, 2013, entitled “KINETIC EXCLUSION AMPLIFICATION OF NUCLEICACID LIBRARIES”;
[0014] U.S. non-provisional patent application No. 13 / 006,206, entitled “DATA PROCESSING SYSTEM AND METHODS”, filed on January 13, 2011;
[0015] U.S. non-provisional patent application No. 14 / 530,299, filed on October 31, 2014, entitled “IMAGE ANALYSIS USEFUL FOR PATTERNEDOBJECTS”;
[0016] U.S. non-provisional patent application No. 15 / 153,953, filed on December 3, 2014, entitled “METHODS AND SYSTEMS FOR ANALYZING IMAGEDATA”;
[0017] U.S. non-provisional patent application No. 14 / 020,570, filed on September 6, 2013, entitled “CENTROID MARKERS FOR IMAGE ANALYSIS OF HIGHDENSITY CLUSTERS IN COMPLEX POLYNUCLEOTIDE SEQUENCING”;
[0018] U.S. non-provisional patent application No. 14 / 530,299, filed on October 31, 2014, entitled “IMAGE ANALYSIS USEFUL FOR PATTERNEDOBJECTS”;
[0019] U.S. non-provisional patent application No. 12 / 565,341, filed on September 23, 2009, entitled “METHOD AND SYSTEM FOR DETERMINING THEACCURACY OF DNA BASE IDENTIFICATIONS”;
[0020] U.S. non-provisional patent application No. 12 / 295,337, filed on March 30, 2007, entitled “SYSTEMS AND DEVICES FOR SEQUENCE BY SYNTHESIS ANALYSIS”;
[0021] U.S. non-provisional patent application No. 12 / 020,739, filed on January 28, 2008, entitled “IMAGE DATA EFFICIENT GENETIC SEQUENCING METHOD AND SYSTEM”;
[0022] U.S. non-provisional patent application No. 13 / 833,619, entitled “BIOSENSORS FOR BIOLOGICAL OR CHEMICAL ANALYSIS AND SYSTEMS AND METHODS FOR SAME”, filed on March 15, 2013 (Attorney’s File No. IP-0626-US).
[0023] U.S. non-provisional patent application No. 15 / 175,489, entitled “BIOSENSORS FOR BIOLOGICAL OR CHEMICAL ANALYSIS AND METHODS OF MANUFACTURING THE SAME”, filed on June 7, 2016 (Attorney’s File No. IP-0689-US).
[0024] U.S. non-provisional patent application No. 13 / 882,088, entitled “MICRODEVICES AND BIOSENSOR CARTRIDGES FOR BIOLOGICAL OR CHEMICAL ANALYSIS AND SYSTEMS AND METHODS FOR THE SAME”, filed on April 26, 2013 (Attorney’s File No. IP-0462-US).
[0025] U.S. non-provisional patent application No. 13 / 624,200, entitled “METHODS AND COMPOSITIONS FOR NUCLEIC ACIDSEQUENCING”, filed on September 21, 2012 (Attorney’s File No. IP-0538-US).
[0026] U.S. Provisional Patent Application No. 62 / 821,602, entitled “Training Data Generation for Artificial Intelligence-Based Sequencing”, filed on March 21, 2019 (Attorney’s File No. ILLM 1008-1 / IP-1693-PRV).
[0027] U.S. Provisional Patent Application No. 62 / 821,618, entitled “Artificial Intelligence-Based Generation of Sequencing Metadata”, filed on March 21, 2019 (Attorney’s File No. ILLM 1008-3 / IP-1741-PRV).
[0028] U.S. Provisional Patent Application No. 62 / 821,681, entitled “Artificial Intelligence-Based Base Calling”, filed on March 21, 2019 (Attorney’s File No. ILLM 1008-4 / IP-1744-PRV).
[0029] U.S. Provisional Patent Application No. 62 / 821,724, entitled “Artificial Intelligence-Based QualityScoring”, filed on March 21, 2019 (Attorney’s File No. ILLM 1008-7 / IP-1747-PRV).
[0030] U.S. Provisional Patent Application No. 62 / 821,766, entitled “Artificial Intelligence-Based Sequencing”, filed on March 21, 2019 (Attorney’s File No. ILLM 1008-9 / IP-1752-PRV).
[0031] Dutch patent application No. 2023310, entitled “Training Data Generation for Artificial Intelligence-Based Sequencing”, filed on June 14, 2019 (Agent's file number ILLM1008-11 / IP-1693-NL).
[0032] Dutch patent application No. 2023311, entitled “Artificial Intelligence-Based Generation of Sequencing Metadata”, filed on June 14, 2019 (Agent’s file number ILLM 1008-12 / IP-1741-NL).
[0033] Dutch patent application No. 2023312 entitled “Artificial Intelligence-Based Base Calling” was filed on June 14, 2019 (Agent’s file number ILLM 1008-13 / IP-1744-NL).
[0034] Dutch patent application No. 2023314, filed on June 14, 2019, entitled "Artificial Intelligence-Based QualityScoring" (Attorney's file number ILLM 1008-14 / IP-1747-NL); and
[0035] Dutch patent application No. 2023316 entitled “Artificial Intelligence-Based Sequencing” was filed on June 14, 2019 (Agent’s file number ILLM 1008-15 / IP-1752-NL).
[0036] U.S. non-provisional patent application No. 16 / 825,987, entitled “Training Data Generation for Artificial Intelligence-Based Sequencing”, filed on March 20, 2020 (Attorney’s File No. ILLM 1008-16 / IP-1693-US).
[0037] U.S. non-provisional patent application No. 16 / 825,991, entitled “Training Data Generation for Artificial Intelligence-Based Sequencing”, filed on March 20, 2020 (Attorney’s File No. ILLM 1008-17 / IP-1741-US).
[0038] U.S. non-provisional patent application No. 16 / 826,126, entitled “Artificial Intelligence-Based Base Calling”, filed on March 20, 2020 (Attorney’s File No. ILLM 1008-18 / IP-1744-US).
[0039] U.S. non-provisional patent application No. 16 / 826,134 entitled “Artificial Intelligence-Based QualityScoring” was filed on March 20, 2020 (Attorney’s File No. ILLM 1008-19 / IP-1747-US).
[0040] U.S. non-provisional patent application No. 16 / 826,168 entitled “Artificial Intelligence-Based Sequencing” was filed on March 21, 2020 (Attorney’s File No. ILLM 1008-20 / IP-1752-PRV).
[0041] U.S. Provisional Patent Application No. 62 / 849,091, filed on May 16, 2019, entitled “Systems and Devices for Characterization and Performance Analysis of Pixel-Based Sequencing” (Attorney’s File No. ILLM 1011-1 / IP-1750-PRV).
[0042] U.S. Provisional Patent Application No. 62 / 849,132, entitled “Base Calling Using Convolutions”, filed on May 16, 2019 (Attorney’s File No. ILLM 1011-2 / IP-1750-PR2).
[0043] U.S. Provisional Patent Application No. 62 / 849,133, entitled “Base Calling Using Compact Convolutions”, filed on May 16, 2019 (Attorney’s File No. ILLM 1011-3 / IP-1750-PR3).
[0044] U.S. Provisional Patent Application No. 62 / 979,384, entitled “Artificial Intelligence-Based Base Calling of Index Sequences”, filed on February 20, 2020 (Attorney’s File No. ILLM 1015-1 / IP-1857-PRV).
[0045] U.S. Provisional Patent Application No. 62 / 979,414, entitled “Artificial Intelligence-Based Many-To-ManyBase Calling”, filed on February 20, 2020 (Attorney’s File No. ILLM 1016-1 / IP-1858-PRV).
[0046] U.S. Provisional Patent Application No. 62 / 979,385, entitled “Knowledge Distillation-Based Compression of Artificial Intelligence-Based Base Caller”, filed on February 20, 2020 (Attorney’s File No. ILLM 1017-1 / IP-1859-PRV).
[0047] U.S. Provisional Patent Application No. 62 / 979,412, entitled “Multi-Cycle Cluster Based Real Time Analysis System”, filed on February 20, 2020 (Attorney’s File No. ILLM 1020-1 / IP-1866-PRV).
[0048] U.S. Provisional Patent Application No. 62 / 979,411 (Attorney’s File No. ILLM 1029-1 / IP-1964-PRV), filed on February 20, 2020, entitled “Data Compression for Artificial Intelligence-Based Base Calling”; and
[0049] U.S. Provisional Patent Application No. 62 / 979,399, entitled “Squeezing Layer for Artificial Intelligence-Based Base Calling”, filed on February 20, 2020 (Attorney’s File No. ILLM 1030-1 / IP-1982-PRV). Background Technology
[0050] The topics discussed in this section should not be considered prior art simply because they are mentioned here. Similarly, problems mentioned in this section or related to the topics provided as background art should not be assumed to have been previously recognized in the prior art. The topics in this section merely represent different methods, which themselves may correspond to specific implementations of the technology protected by the claims.
[0051] Various protocols in biological or chemical research involve conducting a large number of controlled reactions on a locally supported surface or within a predefined reaction chamber. The desired reaction can then be observed or detected, and subsequent analysis can help identify or reveal the properties of the chemicals involved in the reaction. For example, in some multiplex assays, an unknown analyte with an identifiable tag (e.g., a fluorescent tag) can be exposed to thousands of known probes under controlled conditions. Each known probe can be placed in a corresponding well of a microplate. Observing any chemical reactions occurring between the known probe and the unknown analyte within the well can help identify or reveal the properties of the analyte. Other examples of such protocols include known DNA sequencing procedures such as sequencing-while-synthesizing or cyclic array sequencing. In cyclic array sequencing, a dense array of DNA features (e.g., template nucleic acids) is sequenced by repeated cycles of enzymatic manipulation. After each cycle, an image can be captured and subsequently analyzed along with other images to determine the sequence of the DNA features.
[0052] As a more concrete example, a known DNA sequencing system uses pyrosequencing and includes a chip with a fusion-bonded fiber optic panel having millions of wells. Individual capture beads containing sstDNA amplified from a clone of the genome of interest are deposited into each well. After the capture beads are deposited into the wells, nucleotides are sequentially added to the wells by flowing a solution containing specific nucleotides along the panel. The environment within the wells allows a nucleotide to be added to the DNA strand if it is complementary to the DNA strand on the corresponding capture bead. Groups of DNA strands are called clusters. The binding of nucleotides to clusters triggers a process that ultimately generates a chemiluminescent light signal. The system includes a CCD camera positioned directly adjacent to the panel and configured to detect the light signal from the DNA clusters in the wells. Subsequent analysis of images taken throughout the pyrosequencing process can determine the sequence of the genome of interest.
[0053] However, the aforementioned pyrosequencing system may have certain limitations, among other things. For example, the fiber optic panel is acid-etched to form millions of micropores. Although these micropores can be approximately spaced apart from each other, it is difficult to know the precise position of one micropore relative to its neighboring micropores. When the CCD camera is positioned directly adjacent to the panel, the micropores are not uniformly distributed along the pixels of the CCD camera, and therefore, the micropores are not aligned with the pixels in a known manner. Spatial crosstalk, which is inter-micropore crosstalk between adjacent micropores, makes it difficult to distinguish the true light signal from the micropore of interest from other unwanted light signals in subsequent analysis. Moreover, fluorescence emission is essentially isotropic. As the density of analytes increases, managing or resolving unwanted light emission (e.g., crosstalk) from neighboring analytes becomes increasingly challenging. Therefore, the data recorded during sequencing cycles must be carefully analyzed.
[0054] Accurate base detection is crucial for high-throughput DNA sequencing and downstream analyses such as read mapping and genome assembly. Spatial crosstalk between adjacent clusters accounts for a large portion of sequencing errors. Therefore, by correcting for spatial crosstalk in cluster strength data, it is possible to reduce DNA sequencing errors and improve the accuracy of base detection. Attached Figure Description
[0055] This patent or patent application document contains at least one color drawing. A published copy of this patent or patent application with a color drawing will be provided by the Patent Office upon request and payment of the necessary fees. Color drawings are also available in pairs via the Supplements tab.
[0056] In the accompanying drawings, similar reference numerals generally refer to similar parts in all different views. Furthermore, the drawings are not necessarily drawn to scale, but rather emphasize the principles of the disclosed technology. In the following description, various embodiments of the technology disclosed in this invention are described with reference to the following drawings, wherein:
[0057] Figure 1A One implementation of generating a lookup table (LUT) / equalizer filter by training an equalizer is shown.
[0058] Figure 1B An implementation is described that uses the LUT / equalizer filter of Figure 1 to attenuate spatial crosstalk from sensor pixels and to perform base detection on clusters using crosstalk-corrected sensor pixels.
[0059] Figure 2 An example of a sequencing image visualized on a flow cell containing a central / point source of at least five clusters / wells.
[0060] Figure 3 Visualized from Figure 2 The sequencing image is extracted into a pixel patch (yellow), such that the center of target cluster 1 (blue) is included in one example of the center pixel of the pixel patch.
[0061] Figure 4 An example of cluster-to-pixel signal visualization is provided.
[0062] Figure 5 An example of cluster-to-pixel signal overlap is visualized.
[0063] Figure 6 An example of cluster signal patterns is visualized.
[0064] Figure 7 Visualized the attenuation from Figure 3 An example of a subpixel LUT mesh with spatial crosstalk of pixel patches.
[0065] Figure 8 The sub-pixel positions based on the cluster / hole centers within a pixel are shown from... Figure 1B Select the LUT / equalizer filter from the LUT library.
[0066] Figure 9 One implementation is shown in which the center of target cluster 1 (blue) is not substantially concentric with the center of the pixel.
[0067] Figure 10 An implementation method is described that interpolates between a set of selected LUTs and generates corresponding LUT weights.
[0068] Figure 11A weight kernel generator is shown that uses the weights computed using LUTs 12, 7, 8, and 13 to generate weight kernels.
[0069] Figure 12 The diagram illustrates an element-wise multiplier that multiplies the interpolated pixel coefficients of the weight kernel element-wise with the intensity values of the pixels in the pixel patch, and then sums the intermediate products of the multiplication to produce the output.
[0070] Figure 13A , Figure 13B , Figure 13C , Figure 13D , Figure 13E and Figure 13F Examples of coefficients for LUTs 12, 7, 8, and 13 are shown.
[0071] Figure 14A An example of a weight kernel is depicted.
[0072] Figure 14B and Figure 14C An example of the weight kernel generation logic is shown, which is used by the weight kernel generator to generate weight kernels from the weights computed from LUTs 12, 7, 8, and 13.
[0073] Figure 15A and Figure 15B It demonstrates how the interpolated pixel coefficients of the weight kernel maximize the signal-to-noise ratio and recover the underlying signal of target cluster 1 from signals corrupted by crosstalk from clusters 2, 3, 4 and 5.
[0074] Figure 16 One implementation of base-wise Gaussian fitting is shown, in which the base-wise intensity target is contained at its center as the ground truth used for error calculation during training.
[0075] Figure 17 It is a computer system that can be used to implement the technology disclosed in this invention.
[0076] Figure 18 An implementation of an adaptive equalization technique that can be used to train an equalizer is shown.
[0077] Figure 19A , Figure 19B-1 , Figure 19B-2 , Figure 19C and Figure 19D Various performance metrics of the technology disclosed in this invention are demonstrated. Detailed Implementation
[0078] The following description will typically refer to specific structural embodiments and methods. It should be understood that the invention is not intended to be limited to the specifically disclosed embodiments and methods, but rather that other features, elements, methods, and embodiments may be used to practice the invention. Preferred embodiments are described to illustrate the invention and not to limit its scope, which is defined by the claims. Those skilled in the art will recognize many equivalent variations relating to the following description.
[0079] Generate lookup table
[0080] Figure 1 illustrates one implementation of generating a lookup table (LUT) (or LUT library) 106 by training an equalizer 104. The equalizer 104 is also referred to herein as an equalizer-based base detector 104. System 100A includes a trainer 114 that trains the equalizer 104 using least squares estimation. Additional details regarding the equalizer and least squares estimation can be found in the appendix contained in this application.
[0081] Sequencing image 102 is generated during a sequencing run performed by sequencing instruments such as Illumina's iSeq, HiSeqX, HiSeq 3000, HiSeq 4000, HiSeq 2500, NovaSeq 6000, NextSeq 550, NextSeq 1000, NextSeq 2000, NextSeqDx, MiSeq, and MiSeqDx. In one embodiment, the Illumina sequencer employs cyclic reversible termination (CRT) chemistry for base detection. This process relies on growing a nascent strand complementary to a template strand containing fluorescently labeled nucleotides, while tracking the emission signal of each newly added nucleotide. The fluorescently labeled nucleotides have a 3' removable block of fluorophore signal anchored to the nucleotide type.
[0082] Sequencing is performed in repeated cycles, each cycle consisting of three steps: (a) elongating the nascent strand by adding fluorescently labeled nucleotides; (b) exciting the fluorophore using one or more lasers of the sequencing instrument's optical system and imaging it through different filters of the optical system to produce a sequencing image; and (c) lysing the fluorophore and removing the 3' block to prepare for the next sequencing cycle. The incorporation and imaging cycles are repeated until a specified number of sequencing cycles are reached, thereby defining the read length. Using this method, each cycle queries a new location along the template strand.
[0083] The immense power of Illumina sequencers stems from their ability to simultaneously execute and sense millions, or even billions, of analytes undergoing CRT reactions (e.g., clusters). A cluster comprises approximately a thousand identical copies of the template strand, but differs in size and shape. Before sequencing runs, clusters from the template strand are grown by bridging or exclusion amplification of the input library. The purpose of amplification and cluster growth is to increase the intensity of the emission signal, as imaging devices cannot reliably sense the fluorophore signal of a single strand. However, the strands within a cluster are physically close together, so the imaging device perceives the cluster of strands as a single point.
[0084] Sequencing occurs in a flow cell (a small slide holding the input strand). The flow cell is connected to an optical system including a microscope imaging system, an excitation laser, and a fluorescence filter. The flow cell comprises multiple chambers called channels. These channels are physically separated from each other and can contain different labeled sequencing libraries that can be distinguished in the absence of sample cross-contamination. In some embodiments, the flow cell includes a patterned surface. A “patterned surface” refers to the arrangement of different regions in or on the exposed layer of a solid support. For example, one or more of these regions may be feature portions containing one or more amplification primers. Feature portions may be separated by gap regions where no amplification primers are present. In some embodiments, the pattern may be an xy format of feature portions arranged in rows and columns. In some embodiments, the pattern may be a repetitive arrangement of feature portions and / or gap regions. In some embodiments, the pattern may be a random arrangement of feature portions and / or gap regions. Exemplary patterned surfaces that can be used in the methods and compositions described herein are described in U.S. Patent Nos. 8,778,849, 9,079,148, 8,778,848, and 2014 / 0243224, each of which is incorporated herein by reference.
[0085] In some embodiments, the flow cell includes an array of holes or recesses in its surface. This can be fabricated using a variety of techniques as commonly known in the art, including but not limited to photolithography, imprinting, molding, and micro-etching. Those skilled in the art will appreciate that the technique used will depend on the composition and shape of the array substrate.
[0086] The features in the patterned surface can be pores (e.g., micropores or nanopores) in a pore array on a solid support of glass, silicon, plastic, or other suitable patterned and covalently linked gels (such as poly(N-(5-azidoacetamidopentyl)acrylamide-co-acrylamide) (PAZAM, see, for example, U.S. Patent Publication No. 2013 / 184796, WO 2016 / 066586, and WO 2015-002813, each of which is incorporated herein by reference in its entirety)). This method produces gel pads for sequencing that are stable during sequencing runs with a large number of cycles. The covalent connection of the polymer to the pores helps maintain the structured features of the gel during various applications and throughout the lifetime of the structured substrate. However, in many embodiments, the gel does not need to be covalently connected to the pores. For example, under certain conditions, silane-free acrylamide (SFA, see, for example, U.S. Patent No. 8,563,477, the entire contents of which are incorporated herein by reference) that is not covalently attached to any part of a structured substrate can be used as a gel material.
[0087] In a particular embodiment, the structured substrate can be fabricated by: patterning a solid carrier material to have pores (e.g., micropores or nanopores), coating the patterned carrier with a gel material (e.g., PAZAM, SFA, or chemically modified variants thereof, such as an azide version of SFA (azido-SFA)), and polishing the gel-coated carrier, for example via chemical or mechanical polishing, thereby retaining the gel in the pores while removing substantially all of the gel from or inactivating substantially all of the gel in the gap regions between the pores on the surface of the structured substrate. Primer nucleic acids can be attached to the gel material. A solution of target nucleic acids (e.g., fragmented human genome) can then be contacted with the polished substrate such that the individual target nucleic acids are seeded into the individual pores via interaction with primers attached to the gel material; however, due to the absence of gel material or the inactivation of the gel material, the target nucleic acids will not occupy the gap regions. The amplification of the target nucleic acids will be confined to the pores because the absence of gel in the gap regions or gel inactivation will prevent the outward migration of the growing nucleic acid colony. The process is manufacturable and scalable, utilizing conventional micron or nanometer manufacturing methods.
[0088] Imaging devices in sequencing instruments (e.g., solid-state imaging devices such as charge-coupled devices (CCDs) or complementary metal-oxide-semiconductor (CMOS) sensors) take snapshots at multiple locations along the channel in a series of non-overlapping regions (called blocks). For example, each channel may contain sixty-four or ninety-six blocks. A block can hold hundreds of thousands to millions of clusters.
[0089] The output of a sequencing run is a sequencing image, each depicting the intensity emission of a cluster and its surrounding background. The sequencing images depict the intensity emission resulting from nucleotide incorporation into the sequence during sequencing. These intensity emissions originate from the associated analyte / cluster and its surrounding background.
[0090] Sequencing images 102 originate from multiple sequencing instruments, sequencing runs, cycles, flow cells, blocks, wells, and clusters. In one embodiment, the sequencing images are processed by equalizer 104 on an imaging channel basis. Each sequencing run produces m images corresponding to m imaging channels in each sequencing cycle. In one embodiment, each image channel corresponds to one of multiple filter wavelength bands. In another embodiment, each imaging channel corresponds to one of multiple imaging events in a sequencing cycle. In yet another embodiment, each imaging channel corresponds to a combination of illumination using a specific laser and imaging through a specific optical filter. In different embodiments such as 4-channel chemistry, 2-channel chemistry, and 1-channel chemistry, m is 4 or 2. In other embodiments, m is 1, 3, or greater than 4.
[0091] In another implementation, the input data is based on pH changes induced by the release of hydrogen ions during molecular elongation. The pH change is detected and converted into a voltage change proportional to the number of bases introduced (e.g., in the case of IonTorrent). In yet another specific implementation, the input data is constructed based on nanopore sensing, which uses a biosensor to measure the interruption of current as an analyte passes through or near its pore opening, while simultaneously determining the type of base. For example, Oxford Nanopore Technology (ONT) sequencing is based on the concept of passing single-stranded DNA (or RNA) through a nanopore membrane with a voltage difference applied across the membrane. The nucleotides present in the pore will affect the pore's resistance, so the current measurement over time indicates the sequence of DNA bases that have passed through the pore. This current signal (referred to as a "squiggle" due to its appearance when plotted) is raw data collected by the ONT sequencer. These measurements are stored as 16-bit integer data acquisition (DAC) values acquired at, for example, a frequency of 4 kHz. At a DNA chain speed of approximately 450 base pairs / second, this yields an average of about nine raw observations per base. The signal is then processed to identify interruptions in the open-hole signal corresponding to each reading. This maximization of the raw signal involves base detection, the process of converting DAC values into a DNA base sequence. In some implementations, the input data includes normalized or scaled DAC values. Additional information regarding non-image-based sequencing data can be found in U.S. Provisional Patent Application No. 62 / 849,132, filed May 16, 2019, entitled “Base Calling Using Convolutions” (Attorney’s File No. ILLM 1011-2 / IP-1750-PR2); U.S. Provisional Patent Application No. 62 / 849,133, filed May 16, 2019, entitled “Base Calling Using Compact Convolutions” (Attorney’s File No. ILLM 1011-3 / IP-1750-PR3); and U.S. Non-Provisional Patent Application No. 16 / 826,168, filed March 21, 2020, entitled “Artificial Intelligence-Based Sequencing” (Attorney’s File No. ILLM 1008-20 / IP-1752-PRV).
[0092] train
[0093] Equalizer 104 generates a library of LUTs (equalizer filters) 106, each LUT having a subpixel resolution. In one implementation, the number of LUTs 106 generated by equalizer 104 for the LUT library depends on the number of subpixels into which, or which, the sensor pixels of sequencing image 102 are divided. For example, if each sensor pixel of sequencing image 102 can be divided into n×n subpixels (e.g., 5×5 subpixels), then equalizer 104 generates n... 2 106 LUTs (e.g., 25 LUTs).
[0094] In one implementation of this training, data from the sequencing images is binned by well pixel location. For example, for a 5×5 LUT, 1 / 25 of the wells are centered in bin (1,1) (e.g., the top left corner of the sensor pixel), 1 / 25 of the wells are in bin (1,2), and so on. The equalizer coefficients for the center bin of each well are determined using least-squares estimates of the data subsets from the wells in each bin. The input to equalizer 104 is the raw sensor pixels of the sequencing images for those bins. The resulting estimated equalizer coefficients are different for each bin.
[0095] Each LUT has multiple coefficients learned from training. In one implementation, the number of coefficients in the LUT corresponds to the number of sensor pixels used for base detection of clusters. For example, if the local grid size of the sensor pixels (image or pixel patch) used for base detection of clusters is p×p (e.g., a 9×9 pixel patch), then each LUT has p 2 81 coefficients (e.g., 81 coefficients).
[0096] Training generates equalizer coefficients configured to blend / combine the intensity values of pixels in a manner that maximizes the signal-to-noise ratio (SNR). These pixels represent intensity emissions from a target cluster used for base detection and intensity emissions from one or more neighboring clusters. The signal with maximized SNR is the intensity emission from the target cluster, while the noise with minimized SNR is the intensity emission from neighboring clusters, i.e., spatial crosstalk, plus some random noise (e.g., to account for background intensity emissions). The equalizer coefficients are used as weights, and the blending / combining involves performing element-wise multiplications between the equalizer coefficients and the pixel intensity values to compute a weighted sum of these pixel intensity values.
[0097] During training, according to one implementation, equalizer 104 learns to maximize the signal-to-noise ratio through least-squares estimation. In the case of least-squares estimation, equalizer 104 is trained to use equalizer coefficients shared by the pixel intensities around the test aperture and the desired output estimate. Least-squares estimation is well-suited for this purpose because its output minimizes the squared error and accounts for the effects of noise amplification.
[0098] When the intensity channel is open, the desired output is the pulse at the aperture location (point source), while when the intensity channel is closed, the desired output is the background level. In some implementations, ground truth base detection 112 is used to generate the desired output. In some implementations, ground truth base detection 112 is modified to specify the per-aperture DC offset, amplification factor, multiclonal degree, and gain offset parameters included in the least squares estimate. In one implementation, during training, the DC offset (i.e., a fixed offset) is calculated as part of the least squares estimate. During inference, the DC offset is added as a bias to each equalizer calculation.
[0099] In one implementation, Illumina's Real-Time Analysis (RTA) base detector (which does not use an equalizer) is used to estimate the desired output. Details of the RTA can be found in U.S. Patent Application No. 13 / 006,206, which is incorporated herein by reference as fully set forth herein. Because the RTA has a low base detection error rate, the RTA base detector is used as the source of the ground truth base detection 112. The base detection error is averaged across many training examples. In another implementation, the ground truth base detection 112 is obtained using aligned genomic data of good quality, as the aligned genomic data can utilize a reference genome and truth information that combines knowledge from multiple sequencing platforms and sequencing runs to average over noise.
[0100] The ground truth base detection 112 consists of base-specific intensity values that reliably represent the intensity distribution of bases A, C, G, and T, respectively. Base detectors (such as RTAs) perform base detection on clusters by processing the sequencing image 102 and generating per-color intensity values / outputs for each base detection. These per-color intensity values can be considered per-base intensity values because, depending on the type of chemical (e.g., a 2-color chemical or a 4-color chemical), the colors are mapped to each of the bases A, C, G, and T. Bases with the closest matching intensity distribution are detected.
[0101] Figure 16 One implementation of base-by-base Gaussian fitting is illustrated, where the fits contain a base-by-base intensity target at their center, which is used as the ground truth for error calculation during training. The base-by-base intensity output, generated by a base detector for a large number of bases detected in the training data (e.g., tens, hundreds, thousands, or millions of bases), is used to generate the base-by-base intensity distribution. Figure 16A graph with four Gaussian clouds is shown, representing the probability distribution of the base-by-base intensity outputs for bases A, C, G, and T. The intensity values at the center of these four Gaussian clouds are used as ground truth intensity targets (assuming ground truth base detections of A, C, G, and T are 112, respectively), and are referred to herein as intensity targets.
[0102] Consider, during training, that the input image data fed to equalizer 104 is annotated with the base "A" for ground truth base detection. Then, the target / desired output of equalizer 104 is... Figure 16 The intensity value at the center of the green cloud, i.e., the intensity target of base A. Similarly, for the ground truth base detection of base "C", the expected output of equalizer 104 is Figure 16 The intensity value at the center of the blue cloud in the image is the intensity target for base C. Therefore, the target or desired output during the training of equalizer 104 is the average intensity of the corresponding bases A, C, G, and T after averaging across the training data. In one implementation, trainer 114 uses least-squares estimation to fit the coefficients of equalizer 104 to minimize the equalizer output error for these intensity targets.
[0103] In one implementation, during training, equalizer 104 applies coefficients from a given lookup table (LUT) to pixels in a sequencing image labeled with a given base. This involves multiplying the coefficients element-wise by the intensity values of the pixels and generating a weighted sum of intensity values, where the coefficients act as weights. This weighted sum then becomes the predicted output of equalizer 104. Then, based on a cost / error function (e.g., sum of squared errors (SSE)), an error (e.g., least squares error, least mean squares error) is calculated between the weighted sum and an intensity target determined for a given base (e.g., the average intensity observed for a given base, from the center of the corresponding Gaussian fit of the intensity). The cost function (such as SSE) is a differentiable function used to estimate the equalizer coefficients using an adaptive method, so we can evaluate the derivative of the error with respect to the coefficients and then use these derivatives to update the coefficients with values that minimize the error. This process is repeated until the updated coefficients no longer reduce the error. In other implementations, batch least squares is used to train equalizer 104.
[0104] In other embodiments, Figure 16 The base-by-base intensity distribution / Gaussian cloud shown can be generated on a "hole-by-hole" basis, and noise can be corrected by adding DC offset, amplification factor, and / or phase adjustment parameters. Thus, depending on the hole location of a particular hole, the target intensity value for that particular hole can be generated using the corresponding base-by-base Gaussian cloud.
[0105] In one implementation, a bias term is added to the dot product that produces the output of equalizer 104. During training, a similar method (i.e., least squares or least mean squares (LMS)) used to learn the equalizer coefficients can be used to estimate the bias parameter. In some implementations, the value of the bias parameter is a constant equal to 1, i.e., a value that does not change with the input pixel intensity. A bias exists for each set of equalizer coefficients. This bias is learned during training and then fixed for use during inference. The learned bias represents the DC offset used in each equalizer calculation during inference, as well as the learned coefficients for each LUT. This bias reflects random noise caused by different cluster sizes, different background intensities, varying stimulus responses, varying focus, varying sensor sensitivity, and varying lens aberrations.
[0106] In other decision-oriented implementations, it is assumed that the output of equalizer 104 is correct for the training objective.
[0107] In another implementation of this training, equalizer 104 generates a single LUT (equalizer filter) for only one bin, and then uses multiple per-bin interpolation filters 108 to generate the remaining equalizer filters for the remaining bins. In this implementation, for each training example, the sensor pixels surrounding each hole are resampled / interpolated into a well-aligned space (i.e., these holes are centered in their respective pixel patches / local grids). The resampled pixels for each example are then consistently aligned across all holes.
[0108] However, in order to apply the single equalizer filter generated by equalizer 104 to a real online system for base detection, we need to preprocess the raw sensor pixels of the sequencing image to return to a well-aligned space, i.e., perform interpolation on the raw pixels around each hole, with the interpolation parameters varying according to the sub-pixel position of a given hole. To avoid this interpolation process, we pre-compute the total response for a given hole sub-pixel position. We compute the well-aligned equalizer input values by interpolating the raw pixel intensity into the well-aligned pixel space. We convolve the interpolated response and the equalizer response together to reduce computation. Since the interpolation filter varies with the sub-pixel hole position, this gives each sub-pixel hole position a different set of equalizer coefficients / equalizer filters, thus generating the remaining LUTs for the remaining bins. Therefore, in this implementation of training, only the coefficients of a single equalizer filter are trained during training, but the pre-computation process generates a library of LUT-based equalizers by combining bin-specific interpolation filters 108 with the single equalizer filter, where the LUT index is the sub-pixel hole position.
[0109] Trainer 114 can train equalizer 104 and generate trained coefficients for LUT 106 using various training techniques. Examples of training techniques include least squares estimation, ordinary least squares, least mean method, and recursive least squares. Least squares techniques adjust the parameters of the function to best fit the dataset, minimizing the sum of squared residuals. Additional details about least squares estimation algorithms can be found here—Least squares, https: / / en.wikipedia.org / w / index.php?title=Least_squares&oldid=951737821 (last accessed: April 28, 2020), which is incorporated herein by reference as fully described in this article. Ordinary least squares is a least squares method used for estimation in linear regression models. Additional details about the Ordinary Least Squares algorithm can be found here: https: / / en.wikipedia.org / w / index.php?title=Ordinary_least_squares&oldid=951770366 (last accessed: April 28, 2020), which is incorporated herein by reference as if fully described in this document. In other implementations, other estimation algorithms and adaptive equalization algorithms can be used to train the equalizer 104.
[0110] Equalizer 104 can be trained in offline mode. In offline mode, according to one implementation, the trained coefficients of LUT 106 are generated using the following batch least squares equalization logic:
[0111] .
[0112] In the above relationships, the LUT coefficients are β_hat, the pixel intensity is X, and the target is y. A DC term is also added to the pixel intensity and the coefficients (e.g., an additional intensity term that is fixed at 1 for all cases). Then, as an example, consider X as a matrix of size 82 (= 9×9 input intensities plus a constant DC term) × the number of training examples in the batch, and Y as the target output for each training example, i.e., each value is the intensity center of the on / off cloud depending on the ground truth of the training example. β_hat is the set of coefficients that minimizes the sum of squared residuals and also has a size of 82 (= 9×9 coefficients plus 1 DC term).
[0113] Equalizer 104 can also be trained in online mode to adjust the coefficients of LUT 106 while the sequencer is running and sequencing runs are cyclical, to track changes in temperature (e.g., optical distortion), focus, chemical properties, machine-specific variations, etc., on a block-by-block or sub-block basis. In online mode, adaptive equalization is used to generate the trained coefficients of LUT 106. The online mode uses least mean squares as the training algorithm, which is in the form of stochastic gradient descent. Additional details about the least mean squares algorithm can be found here—Least mean squaresfilter, https: / / en.wikipedia.org / w / index.php?title=Least_mean_squares_filter&oldid=941899198 (last accessed: April 28, 2020), which is incorporated herein by reference as fully described in this article.
[0114] The least mean square technique uses the gradient of the squared error relative to each coefficient to move these coefficients in a direction that minimizes a cost function, where the cost function is the expected value of the squared error. This has extremely low computational cost, performing only multiplication and accumulation operations for each coefficient. No long-term storage is required other than the coefficients. The least mean square technique is well-suited for processing massive amounts of data (e.g., parallel processing of data from billions of clusters). Extensions of the least mean square technique include normalized least mean square and frequency domain least mean square, which are also used in this paper. In some implementations, the least mean square technique can be applied in a decision-oriented manner, where it is assumed that our decision is correct, i.e., our error rate is very low, and a small Mu value will filter out any distorted updates due to incorrect base detection.
[0115] Figure 18 An implementation of an adaptive equalization technique that can be used to train equalizer 104 is shown. Here, the equalization logic is y = xh + d, where x is the input pixel intensity, h is the equalizer coefficient, and d is the DC offset. In one implementation, x and h are row and column vectors of length 81, respectively. This vector model is equivalent to the dot product of a 9×9 matrix representing the input pixels and coefficients. The cost is the expected value of the squared error. Gradient updates move each coefficient in a direction that reduces the expected value of the squared error. This results in the following updates:
[0116]
[0117] For most systems, the expectation function It must be approximate. This can be accomplished using the following unbiased estimator.
[0118]
[0119] Where N indicates the number of samples used for this estimation. The simplest case is N = 1.
[0120]
[0121] For this simple case, the update algorithm is as follows:
[0122]
[0123] In fact, this constitutes the update algorithm for the LMS filter.
[0124] In the above relationship, h is a vector of equalizer coefficients (e.g., 9×9 equalizer coefficients), x is a vector of equalizer input intensity (e.g., 9×9 pixels in a pixel patch), and e is the error calculated by the equalizer using 81 values in x, that is, each equalizer output has only 1 error term.
[0125] This update generates new estimates for 9×9 equalizer coefficients, which shift these coefficients in a direction that (on average) reduces the mean squared error (MSE). There are 81 updates, one for each equalizer coefficient. In some implementations, Mu is a small constant used to change the adaptive rate / convergence speed. The DC term update can be calculated in a similar manner. The gain term update can also be calculated in a similar manner.
[0126] The coefficient set can be shared, for example, between blocks, block regions, or flow pool surfaces. This is achieved by saving and restoring the coefficient set as the input data changes.
[0127] In some implementations, these updates are applied slightly differently because linear interpolation is applied to the set of coefficients:
[0128] h(q, n+1) = h(q, n) + λ_q. mu.x(n). e(n)
[0129] In the above relation, h(q, n) is the weight q at loop n, λ_q is the linear interpolation weight of a specific set of coefficients, and due to two-dimensional linear interpolation, it can include four updates of each equalizer output.
[0130] Recursive least squares extends least squares techniques to recursive algorithms. Additional details about recursive least squares algorithms can be found here—Recursive least squares filter, https: / / en.wikipedia.org / w / index.php?title=Recursive_least_squares_filter&oldid=916406502 (last accessed: April 28, 2020), which is incorporated here by reference as it is fully described in this article.
[0131] In a multi-domain implementation, LUT 106 and its trained coefficients can be generated along multiple domains. Examples of these domains include sequencers or sequencing instruments / machines (e.g., Illumina's NextSeq, MiSeq, HiSeq, and their respective models), sequencing protocols and chemicals (e.g., bridging amplification, exclusion amplification), sequencing runs (e.g., forward and reverse), sequencing illumination (e.g., structured, unstructured, angled), sequencing devices (e.g., top-mounted CCD camera, bottom-mounted CMOS sensor, one laser, multiple lasers), imaging techniques (single-channel, dual-channel, four-channel), flow cells (e.g., patterned, unpatterned, embedded on a CMOS chip, bottom-mounted CCD camera), and spatial resolution on the flow cell (e.g., Different regions or quadrants within the flow cell (e.g., different blocks on the flow cell (e.g., edge holes on blocks closer to the laser, camera, or fluid system)) and different regions within a block (e.g., different channels on the block (e.g., edge holes on channels closer to the laser, camera, or fluid system)). Those skilled in the art will recognize that similarly, other optional domains and parameters typically associated with sequencing are included (e.g., image processing algorithms, image registration algorithms, ground truth annotation schemes (e.g., continuous labels such as intensity values, hard labels such as one-hot encoding, soft labels such as softmax scoring), temperature, focus, lens, sequencing reagents, and sequencing buffer).
[0132] Sequencing images generated using the corresponding domains from these domains can be used to create discrete and distinct training sets for the corresponding domains. These discrete training sets can be used to train equalizer 104 to generate LUTs with trained coefficients for the corresponding domains. Depending on which domain or combination of domains is used in the current or ongoing sequencing operation, specially trained coefficients generated for the corresponding domains from the multiple said domains can be stored and accessed accordingly during online mode. For example, for a sequencing operation, a first set of coefficients more suitable for the edge wells of the flow cell and a second set of coefficients more suitable for the center wells of the same flow cell can be used.
[0133] In one implementation, the configuration file can specify different combinations of these domains and can be analyzed during online mode to select different sets of coefficients specific to the domains identified by the configuration file.
[0134] In a multi-training implementation, the equalizer 104 undergoes pre-training and training. That is, the LUT 106 and its coefficients are first trained using a first training technique during the pre-training phase, and then the LUT 106 and its coefficients are retrained or further trained using a second training technique during a subsequent training phase. The first and second training techniques can be any of the training techniques listed above. The first and second training techniques can be the same, or they can be different. For example, the pre-training phase can be an offline mode using batch ordinary least squares training, while the training phase can be an online mode using iterative stochastic least mean squares.
[0135] In some implementations, multi-domain and multi-training implementations can be combined such that domain-specific coefficients are pre-trained and then further trained in a domain-specific manner. That is, further training (e.g., online mode) retrains the domain-specific coefficients using only data representing the domain and similar to the data used in the pre-training phase. In other knowledge transfer implementations, pre-training and training can use training data from across domains; for example, the coefficient set is generated during pre-training using images from patterned flow pooling, but retrained during subsequent training phases using images from unpatterned flow pooling.
[0136] Spatial crosstalk attenuator
[0137] Figure 2 An embodiment is depicted that uses the trained LUT / equalizer filter 106 of Figure 1 to attenuate spatial crosstalk from sensor pixels and to perform base detection on clusters using crosstalk-corrected sensor pixels. The trained equalizer base detector 104 operates during the inference phase when base detection occurs. In some embodiments, Figure 2 The actions shown are performed in the preprocessing stage before the base detection stage and generate crosstalk-corrected image data for base detection by the base detector.
[0138] In one implementation, equalizer coefficients are applied to pixel patches 120 (image patches or local grids of sensor pixels) extracted from sequencing image 116 based on imaging channel basis and target cluster basis. Regarding the imaging channel basis, in some implementations, each sequencing image has image data for multiple imaging channels. Consider the optical system of an Illumina sequencer, which uses two distinct imaging channels: a red channel and a green channel. Then, in each sequencing cycle, this optical system produces a red image with red channel intensity and a green image with green channel intensity, which together form a single sequencing image (like the RGB channels of a typical color image).
[0139] During training, the coefficients are trained / configured to maximize the signal-to-noise ratio (SNR) by minimizing the error between the predicted / estimated output and the expected / actual output. An example of this error is the mean squared error (MSE) or mean squared deviation (MSD). The signal whose SNR is maximized is the intensity emission from the target cluster (e.g., the cluster centered in the image patch) for base detection, while the noise whose SNR is minimized is the intensity emission from one or more neighboring clusters, i.e., spatial crosstalk, plus other noise sources (e.g., to account for background intensity emission). The trained coefficients are element-wise multiplied by the pixels of the image patch to compute a weighted sum of the intensity values of those pixels. This weighted sum is then used for base detection of the target cluster.
[0140] In one embodiment, patch extractor 118 extracts a red pixel patch from the red channel and a green pixel patch from the green channel from a single sequencing image. In other embodiments, a red pixel patch is extracted from the red sequencing image of the test sequencing cycle, and a green pixel patch is extracted from the green sequencing image of the test sequencing cycle. Coefficients of LUT 106 are used to generate a red-weighted sum of the red pixel patches and a green-weighted sum of the green pixel patches. Both the red-weighted sum and the green-weighted sum are then used for base detection of the target cluster. The image patch 120 has a size of w×h, where w (width) and h (height) are any numbers in the range of 1 to 10,000 (e.g., 3×3, 5×5, 7×7, 9×9, 15×15, 25×25). In some embodiments, w and h are the same. In other embodiments, w and h are different. Those skilled in the art will know that data from one, two, three, four or more channels or images can be generated for each sequencing cycle of the target cluster, and one, two, three, four or more patches can be extracted to generate one, two, three, four or more weighted sums for base detection of the target cluster.
[0141] Regarding the target cluster basis for extracting pixel patches 120 from sequencing image 116, pixel extractor 118 extracts pixel patches 120 based on the position of the centers of these clusters / wells on sequencing image 116, such that the center pixel of each extracted pixel patch contains the center of the target cluster / well. In some embodiments, patch extractor 118 locates the cluster / well centers on the sequencing image, identifies those pixels in the sequencing image that contain the cluster / well centers (i.e., center pixels), and extracts pixel patches from the neighborhood of consecutive adjacent pixels surrounding the center pixels.
[0142] Figure 2 An example of a sequencing image 200 is visualized, containing a central / point source of at least five clusters / wells on a flow cell. The pixels of sequencing image 200 depict the intensity emission from target cluster 1 (blue), as well as the intensity emission from additional adjacent clusters 2 (purple), 3 (orange), 4 (brown), and 5 (green).
[0143] Figure 3 A visualization shows an example of pixel patch 300 (yellow) extracted from sequencing image 200, such that the center of target cluster 1 (blue) is contained in the center pixel 206 of pixel patch 300. Figure 3 Other pixels 202, 204, 214 and 216 are also shown, which contain the centers of adjacent clusters 2 (purple), 3 (orange), 4 (brown) and 5 (green), respectively.
[0144] Figure 4 An example of cluster-to-pixel signal 400 is visualized. In one embodiment, the sensor pixel (yellow) is located in the pixel plane. Spatial crosstalk is caused by periodically distributed clusters 412 in the sample plane (e.g., flow cell). In one embodiment, the target cluster and additional adjacent clusters are periodically distributed in a rhomboid shape on the flow cell and fixed to the orifice of the flow cell. In another embodiment, the target cluster and additional adjacent clusters are periodically distributed in a hexagonal shape on the flow cell and fixed to the orifice of the flow cell. The signal cone 402 from this cluster is optically coupled to a local grid of the sensor pixel (e.g., pixel patch 300) through at least one lens (e.g., one or more lenses of a top-mounted or adjacent CCD camera).
[0145] Besides rhomboid and hexagonal shapes, these clusters can also be arranged in other regular shapes (such as squares, rhombuses, triangles, etc.). In other embodiments, these clusters are arranged in a random, non-periodic arrangement on the sample plane. Those skilled in the art will appreciate that these clusters can be arranged in any configuration on the sample plane as required by a particular sequencing implementation.
[0146] Figure 5An example of cluster-to-pixel signal overlap 500 is visualized. Signal cones 402 overlap and strike the sensor pixels, resulting in spatial crosstalk 502.
[0147] Figure 6 An example of cluster signal pattern 600 is visualized. In one implementation, cluster signal pattern 600 follows attenuation pattern 602, wherein the cluster signal is strongest at the cluster center and attenuates as it propagates away from the cluster center.
[0148] Figure 6 An example of equalizer coefficients 604 is also shown, which are trained / configured to maximize the signal-to-noise ratio by calculating a weighted sum of intensity emissions from target cluster 1 and intensity emissions from neighboring clusters 2, 3, 4, and 5. Equalizer coefficients 604 act as weights. This weighted sum is calculated by element-wise multiplying a first matrix including equalizer coefficients 604 with a second matrix including pixel intensity values, where each pixel intensity value is the sum of emissions from one or more of clusters 1, 2, 3, 4, and 5 plus other noise sources measured by pixel sensors in the system.
[0149] Figure 7 An example of a subpixel LUT mesh 700 used to attenuate spatial crosstalk from pixel patch 300 is visualized. Each pixel in pixel patch 300 can be divided into multiple subpixels. Figure 7 In this model, pixel 206, which contains the center of target cluster 1 (blue), is divided into the same number of sub-pixels as the number of trained LUTs 106. That is, pixel 206 is divided into the same number of sub-pixels as the number of bins for which the equalizer 104 generates LUTs 106 during training. Therefore, each sub-pixel of pixel 206 corresponds to a corresponding LUT in a LUT library generated by the equalizer 104 using decision-oriented feedback and least-squares estimation.
[0150] exist Figure 7In the example shown, pixel 206 (the center pixel) is divided into 5×5 sub-pixel LUT grids 700 to produce 25 sub-pixels, each corresponding to one of the 25 LUTs (equalizer filters) generated by the adaptive filter 104 as a training result. Each of these 25 LUTs includes coefficients configured to blend / combine the intensity values of pixels in pixel patch 300 in a manner that maximizes the signal-to-noise ratio (SNR). These pixels depict intensity emissions from target cluster 1 as well as intensity emissions from neighboring clusters 2, 3, 4, and 5. The signal with maximized SNR is the intensity emission from the target cluster, while the noise with minimized SNR is the intensity emission from neighboring clusters 2, 3, 4, and 5, i.e., spatial crosstalk, plus some random noise (e.g., to account for background intensity emissions). These LUT coefficients are used as weights, and the blending / combining involves performing element-wise multiplications between the LUT coefficients and the intensity values of pixels in pixel patch 300 to compute a weighted sum of the intensity values of these pixels.
[0151] The number of coefficients in each of these 25 LUTs is the same as the number of pixels in the pixel patch 300; that is, for 9×9 pixels in the pixel patch 300, each LUT has a 9×9 coefficient grid. This occurs because these coefficients are multiplied element-wise with the pixels in the pixel patch 300.
[0152] In one implementation, a pixel-to-subpixel converter ( Figure 1B (Not shown) Pixel 206 is divided into subpixel LUT grid 700 based on a preset pixel divisor parameter (e.g., 1 / 5 of a pixel per subpixel to generate a 5×5 subpixel LUT grid 700). For example, the pixel can be divided into five subpixel bins with the following boundaries: -0.5, -0.3, -0.1, 0.1, 0.3, 0.5.
[0153] exist Figure 7It should be noted that the center of target cluster 1 (blue) is substantially concentric with the center of the transformed pixel 702. This occurs because the sequencing image 200 and therefore the pixel patch 300 are resampled to ensure that the center of target cluster 1 (blue) is substantially concentric with the center of the transformed pixel 702 by: (i) registering the sequencing image 200 relative to the template image and determining the affine transformation parameters and the nonlinear transformation parameters; (ii) transforming the position coordinates of target cluster 1 (blue) to the image coordinates of the sequencing image 200 using these parameters; and (iii) applying interpolation using the transformed position coordinates of target cluster 1 (blue) to ensure that its center is substantially concentric with the center of the transformed pixel 702. The positions of the holes in the sample plane are known and can be used to calculate the position of the equalizer input for a specific hole in the original pixel space. We can then use interpolation to recover the intensity at those positions from the original image.
[0154] Figure 8 The diagram illustrates the selection of a LUT / equalizer filter from a LUT library 106 based on the sub-pixel position of the cluster / hole center within a pixel. Since the center of the target cluster (blue) falls within a specific sub-pixel 12 of the sub-pixel LUT grid 700, and the specific sub-pixel 12 of pixel 206 corresponds to LUT 12 in LUT library 106, LUT selector 122 selects LUT 12 and its coefficients from LUT library 106 to apply to the pixels of pixel patch 300. Element-wise multiplier 134 then multiplies the coefficients of LUT 12 element-wise by the intensity values of the pixels in pixel patch 300 and sums the products of the multiplications to produce an output (e.g., a weighted sum 136). This output is used for base detection of the target cluster 1 (e.g., by feeding this output as input to base detector 138).
[0155] When the target cluster is substantially concentric with the center of the pixel, as mentioned above... Figure 7 and Figure 8 As discussed, equalizer 104 implements the following equalization logic:
[0156]
[0157] In the above relationships, the hole center coordinates (m, n) are integers to ensure that the hole is substantially aligned with the pixel; p(i,j) is the pixel intensity at position i, j; w(i, j) is the equalizer weight of the pixel at position i, j; i, j are summation constraints that act on the pixel range around the hole centered at p(m, n), for example, -4 <= i <= 4, -4 <= j <= 4; and the output is a weighted average of the input pixels.
[0158] Figure 9One implementation is shown in which the center of target cluster 1 (blue) is not substantially concentric with the center of pixel 206 because no actions such as those concerning... Figure 8 The resampling discussed. In this implementation, the interpolation occurs in a set of selected LUTs 124 to produce an interpolated LUT with interpolation coefficients. The interpolated LUT with interpolation coefficients is also referred to herein as weight kernel 132.
[0159] First of all, like in Figure 8 Therefore, the first LUT, LUT 12, is selected corresponding to the specific sub-pixel whose center falls within target cluster 1 (blue). Then, LUT selector 122 selects an additional sub-pixel lookup table from the sub-pixel lookup table library 106, corresponding to the most contiguous neighboring sub-pixel of the specific sub-pixel. Figure 9 In the LUT library, the nearest consecutive adjacent sub-pixels of a specific sub-pixel 12 are sub-pixels 7, 8 and 13, so LUTs 7, 8 and 13 are selected from LUT library 106 respectively.
[0160] Figure 10 An implementation is described that interpolates between a set of selected LUTs and generates corresponding LUT weights. Interpolator 126 is configured to have interpolation logic (e.g., linear, bilinear, or bicubic interpolation) that uses the coefficients of the selected LUTs 12, 7, 8, and 13 and generates weights 128 for each of LUTs 12, 7, 8, and 13.
[0161] Figure 13A , Figure 13B , Figure 13C , Figure 13D , Figure 13E and Figure 13F Examples of coefficients for LUTs 12, 7, 8, and 13 are shown. These graphs also show examples 1312, 1322, and 1332 of the interpolation logic used by interpolator 126 to calculate the weights 128 for LUTs 12, 7, 8, and 13. These graphs also show examples of the weights 128 calculated for LUTs 12, 7, 8, and 13. These graphs are snapshots of Excel spreadsheets, and the blue arrows and color coding in these graphs are generated by Excel's tracking priority feature to illustrate the interpolation logic.
[0162] Figure 11 A weight kernel generator 130 is shown that uses weights 128 calculated using LUTs 12, 7, 8 and 13 to generate weight kernel 132. Figure 14A An example of weight kernel 132 is depicted. Figure 14B and Figure 14CAn example 1402 illustrates the weight kernel generation logic used by weight kernel generator 130 to generate weight kernel 132 from weights 128 computed from LUTs 12, 7, 8, and 13. Weight kernel 132 includes interpolated pixel coefficients 1412 configured to blend / combine the intensity values of pixels in pixel patch 300 in a manner maximizing the signal-to-noise ratio (SNR), representing intensity emissions from target cluster 1 as well as from neighboring clusters 2, 3, 4, and 5. The signal with maximized SNR is the intensity emission from the target cluster, while the noise with minimized SNR is the intensity emission from neighboring clusters 2, 3, 4, and 5, i.e., spatial crosstalk, plus some random noise (e.g., to account for background intensity emissions). These interpolated pixel coefficients 1412 are used as weights, and the blending / combining involves performing element-wise multiplications between the LUT coefficients and the intensity values of pixels in pixel patch 300 to compute a weighted sum of these intensity values.
[0163] Figure 12 The diagram illustrates an element-wise multiplier 134 that element-wise multiplies the interpolated pixel coefficients 1412 of the weighted kernel 132 with the intensity values of the pixels in the pixel patch 300, and then sums the intermediate products 1202 of the multiplication to produce a weighted sum 136. For each aperture, the optical system operates on a point spread function (the response of the optical system) over the point source (cluster intensity in the aperture). In some implementations, biases are added to this operation to account for noise caused by different cluster sizes, different background intensities, varying stimulus responses, varying focus, varying sensor sensitivity, and varying lens aberrations. The captured image is a superposition of responses from all apertures. The selected LUT equalizes the system response around each aperture to estimate the intensity of the point source from that aperture; that is, it processes the PSF intensity over the local neighborhood / grid of the sensor pixel to estimate the intensity of the point source in the local grid that generates the sensor pixel. This equalizer operation is a dot product of the sensor pixel in the local grid with the equalizer coefficients.
[0164] When the target cluster is not substantially concentric with the center pixel, as mentioned above... Figure 9 , Figure 10 , Figure 11 and Figure 12 As discussed, equalizer 104 implements the following equalization logic. When the aperture is not in the center of a pixel, the output of equalizer 104 is calculated as a function of the virtual pixel intensity p'(i, j) derived from the actual pixel intensity of the sequenced image:
[0165] (1)
[0166] In the above relationships, the hole center coordinates (m, n) can have a fractional part. Each “virtual” equalizer input p’(i, j) is generated by applying an interpolation filter to the pixel neighborhood. In one implementation, a windowed Sinc low-pass filter h(x, y) is used for interpolation. In other implementations, other filters, such as bilinear interpolation filters, can be used.
[0167] The virtual pixel at position (i, j) is calculated using an interpolation filter, as follows:
[0168] (2)
[0169] By combining equations (1) and (2), equalizer 104 uses only the following raw pixel intensities:
[0170]
[0171] In the above relationships, h is fixed given the sub-pixel offsets frac(m) and frac(n); u and v specify the range of pixels used for interpolation to generate the equalizer input; and i and j specify the range of virtual pixels used as the input to equalizer 104.
[0172] For a given sub-pixel offset, only the input pixel changes, not the filter or weights. Therefore, for the center of the sub-pixel offset in each bin, we calculate a fixed set of interpolation equalizer coefficients. The output is then:
[0173]
[0174] In the above relation, Represents the LUT equalizer coefficients for holes with fractional subpixel offsets fm and fn, where (fm, fn) are LUT indices.
[0175] Figure 15A and Figure 15B This demonstrates how the interpolation pixel coefficient 1412 of the weight kernel maximizes the signal-to-noise ratio and recovers the underlying signal of target cluster 1 from signals corrupted by crosstalk from clusters 2, 3, 4 and 5.
[0176] The weighted sum 136 is fed as input to the base detector 138 to produce a base detection 140. The base detector 138 may be a non-neural network-based base detector or a neural network-based base detector, examples of which are described in the patent applications incorporated herein by reference, such as U.S. Patent Applications Nos. 62 / 821,766 and 16 / 826,168.
[0177] In other implementations, the need for interpolation is eliminated by having a large LUT, wherein each LUT has a large number of subpixel bins (e.g., each LUT has 50, 75, 100, 150, 200, 300, etc.).
[0178] Figure 19A The graph representing the base detection error rate is shown using images from the NovaSeq sequencer. The error rate is represented by the cycle on the X-axis. 0.004 on the Y-axis represents a base detection error rate of 0.4%. The error rate here is calculated after mapping the reads to and aligning them to the Phi-X reference (which is a set of high-confidence ground truth values). The blue line represents the legacy base detector. The red line represents the base detector 104 based on the improved equalizer disclosed in this paper. At the cost of limited additional computation, the overall error rate is reduced by 57%. The base error rate is higher in subsequent cycles due to additional noise in the system (e.g., pre-phase / phase stabilization, cluster darkening). The increased performance gain in subsequent cycles is valuable because it shows that we can support longer reads. The performance variation between cycles is also significantly reduced.
[0179] Figure 19B-1 and Figure 19B-2 This paper presents another example of the performance results of the equalizer-based base detector 104 disclosed in this invention on sequencing data from NovaSeq and Vega sequencers. For the NovaSeq sequencer, the equalizer-based base detector 104 disclosed in this invention reduces the base detection error rate by more than 50%. For the Vega sequencer, the equalizer-based base detector 104 disclosed in this invention reduces the base detection error rate by more than 35%.
[0180] Figure 19C This paper presents another example of the performance results of the equalizer-based base detector 104 disclosed in this invention with sequencing data from a NextSeq 2000 sequencer. For the NextSeq 2000 sequencer, the equalizer-based base detector 104 disclosed in this invention reduces the base detection error rate by an average of 10% without affecting throughput.
[0181] Figure 19D An embodiment of the computational resources required for the equalizer-based base detector 104 disclosed in this invention is shown. As shown, the equalizer-based base detector 104 disclosed in this invention can be run using a small number of CPU threads, ranging from two to seven threads. Therefore, the equalizer-based base detector 104 disclosed in this invention is a computationally efficient base detector that significantly reduces the base error rate, and can thus be integrated into most existing sequencers without requiring any additional computational resources or dedicated processors (such as GPUs, FPGAs, ASICs, etc.).
[0182] In this patent application, the terms “cluster,” “well,” “sample,” and “fluorescent sample” are used interchangeably, as a well contains the corresponding cluster / sample / fluorescent sample. As defined herein, “sample” and its derivatives are used in their broadest sense, including any specimen, culture, etc., suspected of containing a target. In some embodiments, a sample includes nucleic acids in the form of DNA, RNA, PNA, LNA, chimeric, or hybrid forms. A sample can include any biological, clinical, surgical, agricultural, atmospheric, or aquatic plant or animal specimen containing one or more nucleic acids. The term also includes any isolated nucleic acid sample, such as genomic DNA, freshly frozen, or formalin-fixed paraffin-embedded nucleic acid specimens. It is also contemplated that the source of a sample can be: a single individual, a collection of nucleic acid samples from genetically related members, nucleic acid samples from genetically unrelated members, a (matched) nucleic acid sample from a single individual (such as a tumor sample and a normal tissue sample), or a sample from a single source containing two different forms of genetic material (such as maternal DNA and fetal DNA obtained from a maternal subject), or a sample containing contaminating bacterial DNA in the presence of plant or animal DNA. In some implementations, the source of nucleic acid material may include nucleic acids obtained from newborns, such as nucleic acids commonly used for newborn screening.
[0183] The nucleic acid sample may include high molecular weight substances, such as genomic DNA (gDNA). The sample may include low molecular weight substances, such as nucleic acid molecules obtained from FFPE samples or archived DNA samples. In another embodiment, the low molecular weight substance includes enzymatically fragmented or mechanically fragmented DNA. The sample may contain cell-free circulating DNA. In some embodiments, the sample may include nucleic acid molecules obtained from biopsy tissue, tumors, scrapings, swabs, blood, mucus, urine, plasma, semen, hair, laser capture microscopy, surgical resection, and other clinical or laboratory samples. In some embodiments, the sample may be an epidemiological sample, an agricultural sample, a forensic sample, or a pathogenic sample. In some embodiments, the sample may include nucleic acid molecules obtained from animals (such as humans or mammalian sources). In another embodiment, the sample may include nucleic acid molecules obtained from non-mammal sources (such as plants, bacteria, viruses, or fungi). In some embodiments, the source of the nucleic acid molecules may be archived or extinct samples or species.
[0184] Additionally, the methods and compositions disclosed herein can be used to amplify nucleic acid samples containing low-quality nucleic acid molecules, such as degraded and / or fragmented genomic DNA from forensic samples. In one embodiment, the forensic sample may include nucleic acids obtained from a crime scene, from a missing persons DNA database, from a laboratory associated with a forensic investigation, or may include forensic samples obtained by law enforcement agencies, one or more military services, or any such personnel. The nucleic acid sample may be a purified sample or a lysate containing crude DNA, such as derived from oral swabs, paper, fabric, or other substrates that can be impregnated with saliva, blood, or other bodily fluids. Thus, in some embodiments, the nucleic acid sample may contain a small amount of DNA (such as genomic DNA) or fragmented portions of DNA. In some embodiments, the target sequence may be present in one or more bodily fluids, including but not limited to blood, sputum, plasma, semen, urine, and serum. In some embodiments, the target sequence may be obtained from the victim's hair, skin, tissue samples, autopsy, or remains. In some embodiments, the nucleic acid including one or more target sequences may be obtained from a deceased animal or human. In some embodiments, the target sequence may include nucleic acids obtained from non-human DNA, such as microbial, plant, or insect DNA. In some embodiments, the target sequence or the amplified target sequence is intended for human identification purposes. In some embodiments, this disclosure relates throughout to methods for identifying characteristics of forensic samples. In some embodiments, this disclosure relates throughout to human identification methods using one or more target-specific primers disclosed herein or designed using one or more target-specific primers with the primer design standards outlined herein. In one embodiment, a forensic sample or human identification sample containing at least one target sequence may be amplified using any one or more target-specific primers disclosed herein or using the primer standards outlined herein.
[0185] As used herein, the term "adjacent" when referring to two reaction sites means that there are no other reaction sites between them. The term "adjacent" can have a similar meaning when referring to adjacent detection paths and adjacent photodetectors (e.g., no other photodetectors between adjacent photodetectors). In some cases, a reaction site may not be adjacent to another reaction site, but may still be within the immediate vicinity of that other reaction site. The first reaction site may be adjacent to the second reaction site when a fluorescence emission signal from the first reaction site is detected by a photodetector associated with the second reaction site. More specifically, the first reaction site may be adjacent to the second reaction site when a photodetector associated with the second reaction site detects, for example, crosstalk from the first reaction site. Adjacent reaction sites may be contiguous, such that they are adjacent to each other; or adjacent sites may be non-contiguous, with a spacing between them.
[0186] Technical improvements and terminology
[0187] All references and similar materials cited in this application, including but not limited to patents, patent applications, articles, books, papers, and web pages, regardless of their format, are expressly incorporated in their entirety by reference. If any of the incorporated references and similar materials differs from or contradicts this application, including but not limited to defined terms, usage of terms, or described techniques, this application shall prevail. Additional information regarding terminology can be found in U.S. Non-Provisional Patent Application No. 16 / 826,168, entitled “Artificial Intelligence-Based Sequencing”, filed March 21, 2020 (Attorney’s File No. ILLM 1008-20 / IP-1752-PRV), and U.S. Provisional Patent Application No. 62 / 821,766, entitled “Artificial Intelligence-Based Sequencing”, filed March 21, 2019 (Attorney’s File No. ILLM 1008-9 / IP-1752-PRV).
[0188] The disclosed technique uses neural networks to improve the quality and quantity of nucleic acid sequence information obtainable from nucleic acid samples, such as nucleic acid templates or their complementary sequences, like DNA or RNA polynucleotides or other nucleic acid samples. Therefore, certain specific implementations of the disclosed technique offer higher throughput polynucleotide sequencing compared to previously available methods, such as higher DNA or RNA sequence data collection rates, higher sequence data collection efficiency, and / or lower cost of obtaining such sequence data.
[0189] The disclosed techniques use neural networks to identify the centers of solid-phase nucleic acid clusters and analyze the light signals generated during sequencing of such clusters to clearly distinguish adjacent, contiguous, or overlapping clusters in order to assign sequencing signals to individual discrete source clusters. Therefore, these and related implementations allow the retrieval of meaningful information, such as sequence data, from regions of high-density cluster arrays where previously unavailable information was obtained due to the confounding effects of overlapping or very closely spaced adjacent clusters, including the effects of overlapping signals emanating from them (e.g., as used in nucleic acid sequencing).
[0190] As described in more detail below, in some embodiments, a composition comprising a solid support having one or more nucleic acid clusters, as provided herein, immobilized thereon is provided. Each cluster contains multiple immobilized nucleic acids of the same sequence and has a recognizable center having a detectable center marker, as provided herein, by which the recognizable center can be distinguished from the immobilized nucleic acids in the surrounding region of the cluster. Methods for manufacturing and using such clusters having recognizable centers are also described herein.
[0191] The specific embodiments disclosed in this invention will be used in many situations where the advantage is gained from the ability to identify, determine, annotate, record, or otherwise assign locations that are substantially central within a cluster, such as high-throughput nucleic acid sequencing, the development of image analysis algorithms for assigning optical or other signals to discrete source clusters, and other applications in which identifying the center of immobilized nucleic acid clusters is desired and beneficial.
[0192] In some specific embodiments, the present invention envisions methods relating to high-throughput nucleic acid analysis, such as nucleic acid sequencing (e.g., "sequencing"). Exemplary high-throughput nucleic acid analyses include, but are not limited to, de novo sequencing, resequencing, whole-genome sequencing, gene expression analysis, gene expression monitoring, epigenetic analysis, genome methylation analysis, allele-specific primer extension (APSE), genetic diversity analysis, whole-genome polymorphism discovery and analysis, single nucleotide polymorphism analysis, hybridization-based sequencing methods, etc. Those skilled in the art will recognize that the methods and compositions of the present invention can be used to analyze a wide variety of different nucleic acids.
[0193] While specific embodiments of the invention have been described in relation to nucleic acid sequencing, they are applicable to any field of analyzing image data acquired at different time points, spatial locations, or other temporal or physical perspectives. For example, the methods and systems described herein can be used in the fields of molecular and cell biology, where image data from microarrays, biological specimens, cells, organisms, etc., is acquired and analyzed at different time points or perspectives. Images can be obtained using any number of techniques known in the art, including but not limited to fluorescence microscopy, optical microscopy, confocal microscopy, optical imaging, magnetic resonance imaging, tomography, etc. As another example, the methods and systems described herein can be applied where image data acquired and analyzed at different time points or perspectives using monitoring, aerial, or satellite imaging techniques, etc. The methods and systems are particularly useful for analyzing images acquired for a field of view, where the observed analyte remains in the same position relative to each other within the field of view. However, the analyte may have different characteristics in individual images; for example, the analyte may appear different in individual images of the field of view. For example, an analyte may appear different in terms of the color of a given analyte detected in different images, the variation in the signal intensity of a given analyte detected in different images, or even the presence of a signal of a given analyte detected in one image and the disappearance of the signal of the analyte detected in another image.
[0194] As used herein, the term "analyte" is intended to refer to a point or region in a pattern that can be distinguished from other points or regions based on its relative position. A single analyte may include one or more specific types of molecules. For example, an analyte may include a single target nucleic acid molecule having a specific sequence, or an analyte may include several nucleic acid molecules having the same sequence (and / or its complementary sequence). Different molecules located at different analytes in the pattern can be distinguished from each other based on the position of the analyte in the pattern. Exemplary analytes include, but are not limited to, holes in a substrate, beads (or other particles) in or on a substrate, protrusions on a substrate, ridges on a substrate, gel material pads on a substrate, or channels in a substrate.
[0195] Any of a variety of target analytes to be detected, characterized, or identified may be used in the devices, systems, or methods described herein. Exemplary analytes include, but are not limited to, nucleic acids (e.g., DNA, RNA, or analogues thereof), proteins, polysaccharides, cells, antibodies, epitopes, receptors, ligands, enzymes (e.g., kinases, phosphatases, or polymerases), small molecule drug candidates, cells, viruses, organisms, etc.
[0196] The terms “analyte,” “nucleic acid,” “nucleic acid molecule,” and “polynucleotide” are used interchangeably herein. In various specific embodiments, nucleic acids may be used as templates (e.g., nucleic acid templates, or complementary nucleic acid sequences to nucleic acid templates) as provided herein for specific types of nucleic acid analysis, including but not limited to nucleic acid amplification, nucleic acid expression analysis, and / or nucleic acid sequencing, or suitable combinations thereof. In some specific embodiments, nucleic acids comprise linear polymers of deoxyribonucleotides in, for example, 3'-5' phosphodiester or other bonds, such as deoxyribonucleic acid (DNA), such as single-stranded and double-stranded DNA, genomic DNA, copy DNA or complementary DNA (cDNA), recombinant DNA, or any form of synthetic or modified DNA. In other embodiments, nucleic acids include linear polymers of ribonucleotides in, for example, 3'-5' phosphodiester or other bonds, such as ribonucleic acid (RNA), such as single-stranded and double-stranded RNA, messenger RNA (mRNA), copy RNA or complementary RNA (cRNA), alternatively spliced mRNA, ribosomal RNA, nucleolar small RNA (snoRNA), microRNA (miRNA), small interfering RNA (sRNA), piwi RNA (piRNA), or any form of synthetic or modified RNA. The length of the nucleic acid in the compositions and methods of the present invention can vary and can be a complete or full-length molecule or fragment or a smaller portion of a larger nucleic acid molecule. In certain embodiments, the nucleic acid may have one or more detectable markers, as described elsewhere herein.
[0197] The terms “analyte,” “cluster,” “nucleic acid cluster,” “nucleic acid population,” and “DNA cluster” are used interchangeably to refer to multiple copies of a nucleic acid template and / or its complementary sequence attached to a solid vector. Typically, and in some preferred embodiments, a nucleic acid cluster comprises multiple copies of a template nucleic acid and / or its complementary sequence, said copies being linked to the solid vector via their 5' ends. The copies of the nucleic acid strands constituting the nucleic acid cluster can be single-stranded or double-stranded. The copies of the nucleic acid template present in the cluster may have nucleotides at corresponding positions that differ from each other, for example, due to the presence of a marker moiety. These corresponding positions may also contain similar structures with different chemical structures but similar Watson-Crick base pairing properties, such as uracil and thymine.
[0198] Nucleic acid populations can also be referred to as “nucleic acid clusters.” Nucleic acid populations can be generated optionally through cluster amplification or bridge amplification techniques, as further detailed elsewhere in this document. Multiple repeats of the target sequence can exist in a single nucleic acid molecule, such as multiplexes generated using rolling circle amplification procedures.
[0199] Depending on the conditions used, the nucleic acid clusters of the present invention can have different shapes, sizes, and densities. For example, the clusters can have substantially circular, polygonal, annular, or ring-shaped shapes. The diameter of the nucleic acid clusters can be designed to be from about 0.2 µm to about 6 µm, from about 0.3 µm to about 4 µm, from about 0.4 µm to about 3 µm, from about 0.5 µm to about 2 µm, from about 0.75 µm to about 1.5 µm, or any diameter in between. In a particular embodiment, the diameter of the nucleic acid clusters is from about 0.5 µm, about 1 µm, about 1.5 µm, about 2 µm, about 2.5 µm, about 3 µm, about 4 µm, about 5 µm, or about 6 µm. The diameter of the nucleic acid clusters can be influenced by several parameters, including but not limited to the number of amplification cycles performed when the clusters are generated, the length of the nucleic acid template, or the density of primers attached to the surface on which the clusters are formed. The density of the nucleic acid clusters can be designed to be typically 0.1 / mm². 2 1 / mm 2 10 / mm 2 100 / mm 2 1,000 / mm 2 10,000 / mm 2 Up to 100,000 / mm 2 Within this range. The invention also partially envisions higher density nucleic acid clusters, such as 100,000 / mm². 2 Up to 1,000,000 / mm 2 and 1,000,000 / mm 2 Up to 10,000,000 / mm 2 .
[0200] As used herein, an analyte is a region of interest within a specimen or field of view. When used in conjunction with microarray devices or other molecular analysis equipment, an analyte refers to a region occupied by similar or identical molecules. For example, an analyte can be an amplified oligonucleotide or any other group of polynucleotides or polypeptides having the same or similar sequences. In other embodiments, an analyte can be any element or group of elements occupying a physical region on a specimen. For example, an analyte can be a piece of land, a body of water, etc. When an analyte is imaged, each analyte has a specific area. Therefore, in many embodiments, an analyte is not just a single pixel.
[0201] The distance between analytes can be described in any number of ways. In some embodiments, the distance between analytes can be described as from the center of one analyte to the center of another. In other embodiments, the distance can be described as from the edge of one analyte to the edge of another, or between the outermost identifiable points of each analyte. The edge of an analyte can be described as a theoretical or actual physical boundary on the chip, or a point within the boundary of the analyte. In other embodiments, the distance can be described relative to a fixed point on the specimen or a fixed point in an image of the specimen.
[0202] Generally, this document will describe several specific embodiments of the analytical methods. It should be understood that systems for performing the methods in an automated or semi-automated manner are also provided. Therefore, this disclosure provides a neural network-based template generation and base detection system, wherein the system may include a processor; a storage device; and a program for image analysis, the program including instructions for performing one or more of the methods described herein. Therefore, the methods described herein can be performed, for example, on a computer having components described herein or known in the art.
[0203] The methods and systems illustrated herein can be used to analyze any of a variety of objects. Particularly useful objects are solid supports or solid-phase surfaces with linked analytes. The methods and systems illustrated herein offer advantages when used with objects having repeating patterns of analytes in the xy-plane. One example is a microarray with a linked set of cells, viruses, nucleic acids, proteins, antibodies, carbohydrates, small molecules (such as drug candidates), bioactive molecules, or other analytes of interest.
[0204] A growing number of applications have been developed for arrays containing analytes such as nucleic acids and peptides. These microarrays typically consist of deoxyribonucleic acid (DNA) or ribonucleic acid (RNA) probes. These probes are specific to nucleotide sequences present in humans and other organisms. In some applications, for example, a single DNA or RNA probe can be attached to a single analyte within the array. Samples, such as those from known humans or organisms, can be exposed to the array, allowing the target nucleic acid (e.g., a gene fragment, mRNA, or its amplicon) to hybridize with a complementary probe at the corresponding analyte in the array. The probes can be labeled in a target-specific process (e.g., by a label present on the target nucleic acid or by an enzyme labeling of the probe or target present in the analyte in a hybridized form). The array can then be examined by scanning light at specific frequencies over the analyte to identify which target nucleic acids are present in the sample.
[0205] Biological microarrays can be used for gene sequencing and similar applications. Generally, gene sequencing involves determining the nucleotide sequence in the length of a target nucleic acid (such as a fragment of DNA or RNA). Relatively short sequences are typically sequenced at each analyte, and the resulting sequence information can be used in various bioinformatics methods to logically fit sequence fragments together, thereby reliably determining sequences of a wider range of genetic material lengths from which they are derived. Automated, computer-based algorithms have been developed for characterizing fragments and have recently been used for genome mapping, gene and function identification, and more. Microarrays are particularly useful for characterizing genome content due to the large number of variants, and this replaces the option of performing numerous experiments on a single probe and target. Microarrays are an ideal form for conducting such research in a practical manner.
[0206] Any of a variety of analyte arrays (also known as “microarrays”) known in the art can be used in the methods or systems described herein. A typical array contains analytes, each with a single probe or probe group. In the latter case, the probe group at each analyte is typically homogeneous, having a single type of probe. For example, in the case of a nucleic acid array, each analyte may have multiple nucleic acid molecules, each having a common sequence. However, in some embodiments, the probe group at each analyte of the array may be heterogeneous. Similarly, a protein array may have analytes containing a single protein or protein group, which typically, but not always, has the same amino acid sequence. Probes can be attached to the surface of the array, for example, through covalent bonding or non-covalent interaction. In some embodiments, probes such as nucleic acid molecules can be attached to the surface via a gel layer, as described, for example, in the following patent applications: U.S. Patent Application Serial No. 13 / 784,368 and U.S. Patent Application Publication No. 2011 / 0059865 A1, each of which is incorporated herein by reference.
[0207] Exemplary arrays include, but are not limited to, BeadChip arrays or other arrays derived from Illumina Corporation (San Diego, Calif.), such as those in which a probe is attached to beads present on a surface (e.g., beads in pores on a surface), such as U.S. Patents 6,266,459, 6,355,431, 6,770,441, 6,859,570, or 7,622,294; or those described in PCT Publication WO 00 / 63437, each of which is incorporated herein by reference. Other examples of commercially available microarrays that may be used include, for example, Affymetrix. ® GeneChip ®Microarrays, or according to what is sometimes called VLSIPS ™ Other microarrays synthesized using ultra-large-scale immobilized polymer synthesis techniques. Dot-like microarrays can also be used in methods or systems according to some specific embodiments of the present invention. Exemplary dot-like microarrays are derived from CodeLink of Amersham Biosciences. ™ Arrays. Another available microarray is the one using inkjet printing methods (such as SurePrint from Agilent Technologies). ™ Microarrays manufactured using [technology].
[0208] Other available arrays include those for nucleic acid sequencing applications. For example, arrays with genomic fragment amplicon (often called clusters) are particularly useful, such as those described in Bentley et al., Nature 456:53-59 (2008); WO 04 / 018497; WO 91 / 06678, WO 07 / 123744; U.S. Patent Nos. 7,329,492, 7,211,414, 7,315,019, 7,405,281, or 7,057,026; or U.S. Patent Application Publication No. 2008 / 0108082 A1, each of which is incorporated herein by reference. Another type of array that can be used for nucleic acid sequencing is an array of particles generated by emulsion PCR technology. Examples are described in Dressman et al., Proc. Natl. Acad. Sci. USA 100:8817-8822 (2003), WO 05 / 010145, U.S. Patent Application Publication 2005 / 0130173, or U.S. Patent Application Publication 2005 / 0064460, each of which is incorporated herein by reference in its entirety.
[0209] Arrays used for nucleic acid sequencing typically have a random spatial pattern of nucleic acid analytes. For example, the HiSeq or MiSeq sequencing platforms from Illumina (San Diego, Calif.) utilize flow cells on which nucleic acid arrays are formed by random seeding followed by bridge amplification. However, patterned arrays can also be used for nucleic acid sequencing or other analytical applications. Exemplary patterned arrays, methods of their fabrication, and methods of use are described in U.S. Patent Application Publication Nos. 13 / 787,396, 13 / 783,043, 13 / 784,368, 2013 / 0116153 A1, and 2012 / 0316086 A1, each of which is incorporated herein by reference. Analytes for such patterned arrays can be used to capture individual nucleic acid template molecules for seeding, followed by, for example, bridge amplification to form a homogeneous population. Such patterned arrays are particularly useful for nucleic acid sequencing applications.
[0210] The size of the analytes on the array (or other objects used in the methods or systems described herein) can be selected to suit a specific application. For example, in some embodiments, the analytes on the array may have a size that accommodates only a single nucleic acid molecule. A surface having multiple analytes within this size range can be used to construct a molecular array for detection at single-molecule resolution. Analytes within this size range can also be used in arrays containing groups of nucleic acid molecules, each of which is present. Thus, the analytes on the array may each have a size no larger than about 1 mm. 2 No larger than approximately 500µm 2 No larger than approximately 100µm 2 No larger than approximately 10µm 2 No larger than approximately 1µm 2 No larger than approximately 500nm 2 or no larger than approximately 100 nm 2 No larger than approximately 10nm 2 No larger than approximately 5nm 2 or no larger than approximately 1 nm 2 The area. Alternatively or otherwise, the analytes in the array will be no less than approximately 1 mm. 2 Not less than approximately 500µm 2 Not less than approximately 100µm 2 Not less than approximately 10µm 2 Not less than approximately 1µm 2 Not less than approximately 500nm 2 Not less than approximately 100nm 2 Not less than approximately 10nm 2 Not less than approximately 5nm 2 or not less than about 1nm 2In practice, analytes can have sizes within a range selected from the upper and lower limits exemplified above. While several size ranges of surface analytes have been illustrated with respect to nucleic acids and their scales, it should be understood that analytes within these size ranges can be used in applications that do not involve nucleic acids. It should also be understood that the size of the analyte is not necessarily limited to the scales used in nucleic acid applications.
[0211] For specific implementations involving objects having multiple analytes (such as arrays of analytes), the analytes can be discrete and spaced apart from each other. Arrays used in the present invention may have analytes spaced by edge-to-edge distances of up to 100 µm, 50 µm, 10 µm, 5 µm, 1 µm, 0.5 µm, or less. Alternatively or additionally, the array may have analytes spaced by edge-to-edge distances of at least 0.5 µm, 1 µm, 5 µm, 10 µm, 50 µm, 100 µm, or greater. These ranges may apply to the average edge-to-edge spacing of the analytes as well as the minimum or maximum spacing.
[0212] In some implementations, the analytes in the array do not need to be discrete; instead, adjacent analytes can be adjacent to each other. Regardless of whether the analytes are discrete, the size of the analytes and / or the spacing between them can vary so that the array can have a desired density. For example, the average analyte spacing in a regular pattern can be at most 100µm, 50µm, 10µm, 5µm, 1µm, 0.5µm, or less. Alternatively or otherwise, the average analyte spacing in a regular pattern can be at least 0.5µm, 1µm, 5µm, 10µm, 50µm, 100µm, or greater. These ranges can also apply to the maximum or minimum spacing of the regular pattern. For example, the maximum analyte spacing in a regular pattern may be up to 100µm, 50µm, 10µm, 5µm, 1µm, 0.5µm or less; and / or the minimum analyte spacing in a regular pattern may be at least 0.5µm, 1µm, 5µm, 10µm, 50µm, 100µm or greater.
[0213] The density of analytes in an array can also be understood as the amount of analyte present per unit area. For example, the average analyte density of an array can be at least approximately 1 × 10⁻⁶. 3 analytes / mm 2 1×10 4 analytes / mm 2 1×10 5 analytes / mm 2 1×10 6 analytes / mm 2 1×10 7 analytes / mm 2 1×10 8 analytes / mm2 Or 1×10 9 analytes / mm 2 Or higher. Alternatively or otherwise, the average analyte density of the array may be up to about 1 × 10⁻⁶. 9 analytes / mm 2 1x10 8 analytes / mm 2 1×10 7 analytes / mm 2 1×10 6 analytes / mm 2 1×10 5 analytes / mm 2 1×10 4 analytes / mm 2 Or 1×10 3 analytes / mm 2 Or lower.
[0214] The above scope may apply to all or part of a regular pattern, including, for example, all or part of an array of analytes.
[0215] The analytes in the pattern can have any of a variety of shapes. For example, when viewed in a two-dimensional plane (such as on the surface of an array), the analytes can appear as circles, rings, ellipses, rectangles, squares, symmetrical shapes, asymmetrical shapes, triangles, polygons, etc. The analytes can be arranged in a regular, repeating pattern, including, for example, hexagonal or linear patterns. The pattern can be selected to achieve the desired fill level. For example, circular analytes are best filled with a hexagonal arrangement. Of course, other fill arrangements can also be used for circular analytes, and vice versa.
[0216] The pattern can be characterized by the number of analytes present in a subset of the smallest geometric unit forming the pattern. This subset may include, for example, at least about 2, 3, 4, 5, 6, 10, or more analytes. Depending on the size and density of the analytes, the geometric unit may occupy less than 1 mm. 2 500µm 2 100µm 2 50µm 2 10µm 2 1µm 2 500nm 2 100nm 2 50nm 2 10nm 2 Or even smaller areas. Alternatively, or otherwise, geometric units can occupy more than 10 nm. 2 50nm 2 100nm 2 500nm2 1µm 2 10µm 2 50µm 2 100µm 2 500µm 2 1mm 2 Or a larger area. The characteristics of the analytes in the geometric unit (such as shape, size, spacing, etc.) can be selected from those described more generally in this paper for analytes in arrays or patterns.
[0217] An array with a regular pattern of analytes can be ordered in terms of their relative positions, but random in terms of one or more other characteristics of each analyte. For example, in the case of a nucleic acid array, the nucleic acid analytes can be ordered in terms of their relative positions, but the knowledge of the sequence of the nucleic acid material present at any given analyte is random. As a more specific example, a nucleic acid array formed by seeding a repeating pattern of analytes with template nucleic acids and amplifying the template at each analyte to form a copy of the template at the analyte (e.g., by cluster amplification or bridge amplification) will have a regular pattern of nucleic acid analytes, but the distribution of the nucleic acid sequence throughout the array will be random. Therefore, typically detecting the presence of nucleic acid material on an array produces a repeating pattern of analytes, while sequence-specific detection produces a non-repetitive distribution of the signal throughout the array.
[0218] It should be understood that the descriptions of patterns, sequences, randomness, etc. in this document refer not only to analytes on objects, such as arrays, but also to analytes in images. Therefore, patterns, sequences, randomness, etc., can exist in any of the various formats used to store, manipulate, or transmit image data, including but not limited to computer-readable media or computer components, such as graphical user interfaces or other output devices.
[0219] As used herein, the term "image" is intended to represent a representation of all or part of an object. This representation can be a reproduction of an optically detected signal. For example, an image can be obtained from fluorescence, emission, scattering, or absorption signals. The portion of an object present in an image can be a surface of the object or another xy-plane. Typically, an image is a two-dimensional representation, but in some cases, the information in an image may derive from three or more dimensions. An image does not need to include the signals from an optically detected signal. Instead, non-optical signals may be present. Images can be provided in computer-readable formats or media, such as one or more of those set forth elsewhere herein.
[0220] As used herein, an “image” is a reproduction or representation of at least a portion of a specimen or other object. In some embodiments, the reproduction is an optical reproduction, for example, produced by a camera or other optical detector. The reproduction can be a non-optical reproduction, such as a representation of an electrical signal obtained from a nanopore analyte array or from an ion-sensitive CMOS detector. In certain embodiments, non-optical reproductions may be excluded from the methods or apparatus described herein. An image may have a resolution capable of distinguishing analytes in a specimen present at any of a variety of intervals, including, for example, intervals less than 100 µm, 50 µm, 10 µm, 5 µm, 1 µm, or 0.5 µm.
[0221] As used herein, terms such as “acquisition” and “collection” refer to any part of the process of obtaining an image file. In some specific implementations, data acquisition may include generating images of a specimen, locating signals in a specimen, instructing detection equipment to locate or generate images of signals, giving instructions for further analysis or conversion of image files, and any number of conversions or manipulations of image files.
[0222] As used herein, the term "template" refers to a representation of the position or relationship between signals or analytes. Thus, in some embodiments, a template is a physical grid having representations of signals corresponding to analytes in a specimen. In some embodiments, a template may be a graph, table, text file, or other computer file indicating the position corresponding to an analyte. In the embodiments presented herein, a template is generated to track the position of analytes in a specimen across an image set of specimens captured at different reference points. For example, a template may be a set of x,y coordinates or a set of values describing the orientation and / or distance of one analyte relative to another.
[0223] As used herein, the term "specimen" can refer to the object or area of an object from which an image is captured. For example, in a specific embodiment of imagery of the Earth's surface, a piece of land can serve as a specimen. In other specific embodiments of biomolecular analysis in a flow cell, the flow cell can be divided into any number of sub-sections, each of which can serve as a specimen. For example, the flow cell can be divided into various flow channels or channels, and each channel can be further divided into 2, 3, 4, 5, 6, 7, 8, 9, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110, 120, 140, 160, 180, 200, 400, 600, 800, 1000, or more individual imaging regions. One example of a flow cell has eight channels, with each channel divided into 120 specimens or blocks. In another specific embodiment, a specimen can consist of multiple blocks or even the entire flow cell. Thus, the image of each specimen can represent an imaging region of a larger surface.
[0224] It should be understood that the references to ranges and ordinal lists in this article include not only the enumerated numbers, but also all real numbers between the enumerated numbers.
[0225] As used herein, a "reference point" refers to any temporal or physical difference between images. In a preferred embodiment, the reference point is a time point. In a more preferred embodiment, the reference point is a time point or cycle during a sequencing reaction. However, the term "reference point" can include other aspects that distinguish or separate images, such as angular aspects, rotational aspects, temporal aspects, or other aspects that distinguish or separate images.
[0226] As used herein, an "image subset" refers to a set of images within a collection. For example, a subset may contain 1, 2, 3, 4, 6, 8, 10, 12, 14, 16, 18, 20, 30, 40, 50, or 60 images, or any number of images selected from an image set. In a particular embodiment, a subset may contain no more than 1, 2, 3, 4, 6, 8, 10, 12, 14, 16, 18, 20, 30, 40, 50, or 60 images, or any number of images selected from an image set. In a preferred embodiment, images are obtained from one or more sequencing cycles, with four images associated with each cycle. Thus, for example, a subset could be a set of 16 images obtained through four cycles.
[0227] The base refers to the following nucleotide bases or nucleotides: A (adenine), C (cytosine), T (thymine), or G (guanine). The terms "base" and "nucleotide" are used interchangeably in this patent application.
[0228] The term "chromosome" refers to the genetic carrier of a living cell, derived from a chain of chromatin containing DNA and protein components (especially histones). This article uses the standard, internationally recognized chromosome numbering system for individual human genomes.
[0229] The term "site" refers to a unique location on a reference genome (e.g., chromosome ID, chromosome position, and orientation). In some implementations, a site can be the location of a residue, sequence tag, or fragment on a sequence. The term "genetic site" can be used to refer to a specific location of a nucleic acid sequence or polymorphism on a reference chromosome.
[0230] The term "sample" as used herein refers to a sample typically derived from a biological fluid, cell, tissue, organ, or organism and containing nucleic acids or mixtures thereof, wherein the nucleic acids or mixtures contain at least one nucleic acid sequence to be sequenced and / or phased. Such samples include, but are not limited to, sputum / oral fluid, amniotic fluid, blood, blood fractions, fine-needle biopsy samples (e.g., surgical biopsy, fine-needle biopsy, etc.), urine, peritoneal fluid, pleural fluid, tissue explants, organ cultures, and any other tissue or cell preparations, fractions or derivatives thereof, or fractions or derivatives isolated therefrom. While samples are typically taken from human subjects (e.g., patients), samples may be taken from any organism possessing chromosomes, including but not limited to dogs, cats, horses, goats, sheep, cattle, pigs, etc. Samples may be used directly as is from their biological source or after pretreatment to alter their properties. For example, such pretreatment may include preparing plasma from blood, diluting viscous fluids, etc. Pretreatment methods may also include, but are not limited to, filtration, precipitation, dilution, distillation, mixing, centrifugation, freezing, lyophilization, concentration, amplification, nucleic acid fragmentation, inactivation of interfering components, addition of reagents, lysis, etc.
[0231] The term "sequence" includes or represents a chain of nucleotides coupled to each other. Nucleotides may be based on DNA or RNA. It should be understood that a sequence may include multiple subsequences. For example, a single sequence (e.g., the sequence of a PCR amplicon) may have 350 nucleotides. A sample read may include multiple subsequences within these 350 nucleotides. For example, a sample read may include a first flanking sequence and a second flanking sequence having, for example, 20-50 nucleotides. The first flanking sequence and the second flanking sequence may be located on either side of a repeating fragment having a corresponding subsequence (e.g., 40-100 nucleotides). Each flanking sequence may include a primer subsequence (e.g., 10-30 nucleotides) (or a portion thereof). For ease of reading, the term "subsequence" will be referred to as "sequence," but it should be understood that two sequences are not necessarily separated from each other on a common strand. To distinguish the various sequences described herein, sequences may be assigned different labels (e.g., target sequence, primer sequence, flanking sequence, reference sequence, etc.). Other terms such as "allele" may be assigned different labels to distinguish similar objects. The terms "read segment" and "sequential read segment" can be used interchangeably in this application.
[0232] The term "paired-end sequencing" refers to a sequencing method that sequences both ends of a target fragment. Paired-end sequencing can help detect genomic rearrangements and repetitive fragments, as well as gene fusions and new transcripts. Methods for paired-end sequencing are described in PCT Publication WO07010252, PCT Application Serial No. PCTGB2007 / 003798, and U.S. Patent Application Publication US 2009 / 0088327, each of which is incorporated herein by reference. In one example, a series of operations can be performed as follows: (a) generating nucleic acid clusters; (b) linearizing the nucleic acids; (c) hybridizing a first sequencing primer and performing repeated cycles of extension, scanning, and unblocking as described above; (d) "reversing" the target nucleic acid on the flow cell surface by synthesizing complementary copies; (e) linearizing the resynthesized strand; and (f) hybridizing a second sequencing primer and performing repeated cycles of extension, scanning, and unblocking as described above. The reversal operation can be performed by delivering reagents as described above for a single cycle of bridge amplification.
[0233] The terms "reference genome" or "reference sequence" refer to any specific known genome sequence of any organism, whether partial or complete, that can be used to reference an identified sequence from a subject. For example, reference genomes for human subjects and many other organisms can be found at the National Center for Biotechnology Information (ncbi.nlm.nih.gov). A "genome" refers to the complete genetic information of an organism or virus expressed as a nucleic acid sequence. A genome includes both genes and non-coding sequences of DNA. A reference sequence can be larger than the read it is aligned to. For example, a reference sequence can be at least about 100 times larger, or at least about 1000 times larger, or at least about 10,000 times larger, or at least about 105 times larger, or at least about 106 times larger, or at least about 107 times larger than the aligned read. In one example, the reference genome sequence is a sequence of the full-length human genome. In another example, the reference genome sequence is limited to a specific human chromosome, such as chromosome 13. In some specific implementations, the reference chromosome is a chromosome sequence from human genome version hg19. Such sequences may be referred to as chromosomal reference sequences, but the term reference genome is intended to encompass them. Other examples of reference sequences include genomes of other species, as well as chromosomes, subchromosomal regions (such as chains), etc., of any species. In various specific implementations, a reference genome is a shared sequence or other combination derived from multiple individuals. However, in some applications, the reference sequence may be taken from a specific individual. In other specific implementations, “genome” also encompasses the so-called “graphic genome,” which uses a specific storage format and representation of the genome sequence. In one specific implementation, a graphical genome stores data in a linear file. In another specific implementation, a graphical genome refers to a representation in which alternative sequences (e.g., different copies of chromosomes with minor differences) are stored as different paths in a graph. Further details on specific implementations of graphical genomes can be found at https: / / www.biorxiv.org / content / biorxiv / early / 2018 / 03 / 20 / 194530.full.pdf, the contents of which are incorporated herein by reference in their entirety.
[0234] The term "read" refers to a collection of sequence data describing a fragment of nucleotide sample or reference. The term "read" can refer to a sample read and / or a reference read. Typically, although not strictly necessary, a read represents a short sequence of adjacent base pairs in a sample or reference. A read can be symbolically represented by the base pair sequence (ATCG form) of the sample or reference fragment. Reads can be stored in a storage device and processed as appropriate to determine whether the read matches a reference sequence or meets other criteria. Reads can be obtained directly from the sequencing device or indirectly from stored sequence information about the sample. In some cases, reads are DNA sequences of sufficient length (e.g., at least about 25 bp) that can be used to identify larger sequences or regions, for example, that can be aligned and specifically assigned to chromosomal or genomic regions or genes.
[0235] Next-generation sequencing methods include, for example, sequencing-by-synthesis (Illumina), pyrosequencing (454), ion semiconductor sequencing (Ion Torrent sequencing), single-molecule real-time sequencing (Pacific Biosciences), and ligation-by-ligation sequencing (SOLiD sequencing). Depending on the sequencing method, the length of each read can vary from approximately 30 bp to over 10,000 bp. For example, DNA sequencing using SOLiD sequencers produces nucleic acid reads of approximately 50 bp. Ion Torrent sequencing, on the other hand, produces nucleic acid reads of up to 400 bp, and 454 pyrosequencing produces nucleic acid reads of approximately 700 bp. Single-molecule real-time sequencing methods can produce reads of 10,000 to 15,000 bp. Therefore, in certain implementations, the length of nucleic acid sequence reads is 30 bp–100 bp, 50 bp–200 bp, or 50 bp–400 bp.
[0236] The terms "sample read," "sample sequence," or "sample fragment" refer to sequence data of a genome sequence of interest from a sample. For example, a sample read contains sequence data from a PCR amplicon with forward and reverse primer sequences. Sequence data can be obtained from any chosen sequencing method. Sample reads can, for example, come from sequencing-by-synthesis (SBS) reactions, sequencing-by-ligation reactions, or any other suitable sequencing method requiring determination of the length and / or identity of repetitive elements. Sample reads can be shared (e.g., averaged or weighted) sequences derived from multiple sample reads. In some specific implementations, providing a reference sequence includes identifying the gene site of interest based on the primer sequences of the PCR amplicon.
[0237] The term "raw fragment" refers to sequence data of a portion of the genome sequence of interest that at least partially overlaps with a specified or secondary position of interest in a sample read or fragment. Non-limiting examples of raw fragments include double-spliced fragments, single-spliced fragments, double-unspliced fragments, and single-unspliced fragments. The term "raw" is used to indicate that a raw fragment includes sequence data that has some relationship to the sequence data in the sample read, regardless of whether the raw fragment exhibits supporting variants corresponding to and verifying or confirming potential variants in the sample read. The term "raw fragment" does not imply that the fragment necessarily includes supporting variants that validate variant detection in the sample read. For example, when a variant detection application determines that a sample read exhibits a first variant, the application may determine that one or more raw fragments lack a corresponding type of "supporting" variant that would have been expected given the variant in the given sample read.
[0238] The terms "mapping" and "alignment" refer to the process of comparing a read or tag with a reference sequence to determine whether the reference sequence contains the read sequence. If the reference sequence contains the read, the read can be mapped to the reference sequence, or, in some implementations, to a specific location within the reference sequence. In some cases, alignment simply indicates whether a read is a member of a particular reference sequence (i.e., whether the read is present in that reference sequence). For example, aligning a read with a reference sequence of human chromosome 13 will indicate whether the read is present in the reference sequence of chromosome 13. The tool that provides this information may be called a set membership tester. In some cases, alignment further indicates the location in the reference sequence to which the read or tag is mapped. For example, if the reference sequence is a human whole genome sequence, alignment may indicate that the read is present on chromosome 13, and may also indicate that the read is present on a specific strand and / or site on chromosome 13.
[0239] The term "insertion deletion" refers to the insertion and / or deletion of bases in an organism's DNA. Microinsertion deletions represent insertions that result in a net change of 1 to 50 nucleotides. In coding regions of the genome, frameshift mutations occur unless the insertion / deletion is a multiple of 3 in length. Insertion deletions can be contrasted with point mutations. An insertion deletion involves inserting and deleting nucleotides from the sequence, while a point mutation is a substitution that replaces one nucleotide without changing the total number of nucleotides in the DNA. Insertion deletions can also be contrasted with tandem base mutations (TBMs), which can be defined as substitutions at adjacent nucleotides (primarily two adjacent nucleotides, but substitutions at three adjacent nucleotides have been observed).
[0240] The term "variation" refers to a nucleic acid sequence that differs from a nucleic acid reference. Typical nucleic acid sequence variations include, but are not limited to, single nucleotide polymorphisms (SNPs), short deletions and insertions (Indels), copy number variations (CNVs), microsatellite markers or short tandem repeats, and structural variations. Somatic variant detection is the work of identifying variations present at low frequencies in a DNA sample. Somatic variant detection is of interest in the context of cancer treatment. Cancer is caused by the accumulation of mutations in DNA. DNA samples from tumors are often heterogeneous, including some normal cells, some cells in the early stages of cancer progression (with fewer mutations), and some late-stage cells (with more mutations). Due to this heterogeneity, when sequencing tumors (e.g., from FFPE samples), somatic mutations will typically appear at low frequencies. For example, SNVs may be seen in only 10% of reads covering a given base. Variations to be classified as somatic or germline by a variant classifier are also referred to herein as "variables to be tested".
[0241] The term "noise" refers to erroneous variant detection caused by one or more errors in the sequencing process and / or variant detection application.
[0242] The term "variation frequency" refers to the relative frequency of an allele (genetic variation) at a specific gene locus within a population, expressed as a fraction or percentage. For example, a fraction or percentage could be the fraction of all chromosomes in the population carrying that allele. By way of example, sample variation frequency represents the relative frequency of an allele / variation at a specific gene locus / position along a genome sequence of interest relative to a "population," which corresponds to the number of reads and / or samples of the genome sequence of interest obtained from an individual. Similarly, baseline variation frequency represents the relative frequency of an allele / variation at a specific gene locus / position along one or more baseline genome sequences, where a "population" corresponds to the number of reads and / or samples of one or more baseline genome sequences obtained from a population of normal individuals.
[0243] The term "variant allele frequency (VAF)" refers to the percentage of sequencing reads with observed matching variants divided by the overall coverage at the target site. VAF is a measure of the proportion of sequencing reads carrying variants.
[0244] The terms "position," "designated position," and "genetic locus" refer to the location or coordinates of one or more nucleotides within a nucleotide sequence. They also refer to the location or coordinates of one or more base pairs within a nucleotide sequence.
[0245] The term "haplotype" refers to a combination of alleles that are inherited together at adjacent loci on a chromosome. A haplotype can be a single gene locus, multiple gene loci, or an entire chromosome, depending on the number of recombination events (if any) that occur between a given set of gene loci.
[0246] The term "threshold" in this document refers to a numeric or non-numeric value used as a cutoff value to characterize a sample, nucleic acid, or a portion thereof (e.g., a read). Thresholds can be varied based on empirical analysis. Thresholds can be compared to measurements or calculated values to determine whether sources producing such values should be classified in a particular manner. Thresholds can be identified based on experience or analysis. The choice of threshold depends on the level of confidence the user expects to have to classify. Thresholds can be selected for specific purposes (e.g., to balance sensitivity and selectivity). As used herein, the term "threshold" indicates a point at which the analytical process can be altered and / or a point at which an action can be triggered. Thresholds do not need to be a predetermined number. Instead, thresholds can be, for example, a function based on multiple factors. Thresholds can be adjusted as needed. Furthermore, thresholds can indicate upper limits, lower limits, or a range between limits.
[0247] In some implementations, a metric or score based on sequencing data may be compared to a threshold. As used herein, the terms "metric" or "score" may include a value or result determined by sequencing data, or may include a function based on a value or result determined by sequencing data. Like thresholds, metrics or scores may be adjusted as appropriate. For example, a metric or score may be a normalized value. As an example of a score or metric, one or more implementations may use a count score when analyzing data. A count score may be based on the number of sample reads. Sample reads may have undergone one or more filtering stages, resulting in sample reads having at least one common characteristic or quality. For example, each sample read used to determine the count score may have been aligned to a reference sequence or may have been assigned as a potential allele. The number of sample reads with common characteristics may be counted to determine the read count. A count score may be based on the read count. In some implementations, a count score may be a value equal to the read count. In other implementations, a count score may be based on the read count and other information. For example, a count score may be based on the read count of a specific allele at a gene locus and the total number of reads at the gene locus. In some implementations, the count score may be based on read counts at gene loci and previously obtained data. In some implementations, the count score may be a normalized score between predetermined values. The count score may also be a function of read counts from other gene loci in the sample or from read counts from other samples running in parallel with the sample of interest. For example, the count score may be a function of read counts for a specific allele and read counts at other gene loci in the sample and / or read counts from other samples. For example, read counts from other gene loci and / or read counts from other samples can be used to normalize the count score for a specific allele.
[0248] The term "coverage" or "fragment coverage" refers to the count or other measure of multiple sample reads of the same segment of a sequence. Read counts can represent the number of reads covering the corresponding segment. Alternatively, coverage can be determined by multiplying the read counts by a specified factor based on historical knowledge, sample knowledge, gene locus knowledge, etc.
[0249] The term "read depth" (usually a number followed by a "×") refers to the number of sequence reads with overlapping alignments at the target site. This is typically expressed as the average or percentage of reads within a set of intervals (such as exons, genes, or groups) that exceed a cutoff value. For example, a clinical report might state that the average group coverage is 1,105×, with 98% of the target bases covered by >100×.
[0250] The term "base detection quality score" or "Q score" refers to a PHRED-scale probability in the range of 0-50, which is inversely proportional to the probability of a single sequenced base being correct. For example, a T base detection with a Q of 20 is considered likely to be correct, with a probability of 99.99%. Any base detection with a Q < 20 should be considered low quality, and any variant identified when a significant proportion of sequencing reads supporting the variant are of low quality should be considered a potential false positive.
[0251] The term "variant read" or "number of variant reads" refers to the number of sequencing reads that support the presence of a variant.
[0252] Regarding "chain" (or DNA chain), the genetic information in DNA can be represented as a string of the letters A, G, C, and T. For example, 5' – AGGACA – 3'. Typically, the sequence is written in the direction shown here, i.e., the 5' end to the left and the 3' end to the right. DNA can sometimes appear as a single-stranded molecule (as in some viruses), but we usually find DNA as a double-stranded unit. It has a double helix structure with two antiparallel strands. In this context, the term "antiparallel" refers to the two strands extending in parallel but with opposite polarities. Double-stranded DNA is held together by base pairing, and the pairing always results in adenine (A) pairing with thymine (T) and cytosine (C) pairing with guanine (G). This pairing is called complementarity, and one strand of DNA is called the complementary sequence of the other strand. Therefore, double-stranded DNA can be represented as two strings like this: 5' – AGGACA – 3' and 3' – TCCTGT – 5'. Note that the two strands have opposite polarities. Therefore, the two DNA strands can be referred to as the reference strand and its complementary strand, the forward strand and the reverse strand, the top strand and the bottom strand, the sense strand and the antisense strand, or the Watson strand and the Crick strand.
[0253] Read alignment (also known as read mapping) is the process of finding the location of a sequence in the genome. Once alignment is performed, the “map quality” or “map quality score (MAPQ)” of a given read quantifies the probability that it is correctly located in the genome. Map quality is encoded on the phred scale, where P is the probability of an incorrect alignment. The probability is calculated as follows: MAPQ stands for Mapping Quality. For example, a Mapping Quality of 40 = 10⁻⁴, meaning there's a 0.01% chance the read will be incorrectly aligned. Therefore, Mapping Quality is associated with several alignment factors, such as the base quality of the read, the complexity of the reference genome, and paired-end information. Regarding the first factor, low base quality means the observed sequence is likely incorrect, resulting in an incorrect alignment. Regarding the second factor, Mappability refers to the complexity of the genome. Repetitive regions are more difficult to map, and reads falling into these regions typically receive low Mapping Quality. In this case, MAPQ reflects the fact that the read is not uniquely aligned, and its true origin cannot be determined. Regarding the third factor, in the case of paired-end sequencing data, consistent pairs are more likely to be good alignments. Higher Mapping Quality generally indicates better alignment. Reads aligned with good Mapping Quality typically mean good read sequences and alignments with minimal mismatches in highly mappable regions. MAPQ values can be used as a quality control for alignment results. The proportion of aligned reads with a MAPQ higher than 20 is often used for downstream analysis.
[0254] As used herein, "signal" refers to a detectable event, such as emission in an image, preferably light emission. Therefore, in a preferred embodiment, a signal may represent any detectable light emission (i.e., a "spot of light") captured in an image. Thus, as used herein, "signal" may refer to actual emission from the analyte of the specimen, and may also refer to stray emission unrelated to the actual analyte. Therefore, a signal may be generated by noise and may subsequently be discarded because it does not represent the actual analyte of the specimen.
[0255] As used herein, the term "cluster" refers to a group of signals. In a particular embodiment, the signals originate from different analytes. In a preferred embodiment, a signal cluster is a group of signals aggregated together. In a more preferred embodiment, a signal cluster represents a physical region covered by a single amplified oligonucleotide. Ideally, each signal cluster should be observed as several signals (one per template cycle, and possibly more due to crosstalk). Therefore, in cases where two (or more) signals are included in a template from the same signal cluster, a duplicate signal is detected.
[0256] As used herein, terms such as “minimum,” “maximum,” “minimize,” “maximize,” and their syntactic variations may include values that are not absolute maximum or minimum values. In some implementations, these values include values close to the maximum and values close to the minimum. In other implementations, these values may include local maximum and / or local minimum values. In some implementations, these values include only absolute maximum or minimum values.
[0257] As used herein, "crosstalk" refers to a signal detected in one image also being detected in a separate image. In a preferred embodiment, crosstalk can occur when emitted signals are detected in two separate detection channels. For example, in the case where the emitted signal appears in one color, the emission spectrum of that signal may overlap with that of another emitted signal in another color. In a preferred embodiment, fluorescent molecules used to indicate the presence of nucleotide bases A, C, G, and T are detected in separate channels. However, because the emission spectra of A and C overlap, some of the C color signal can be detected during detection using the A color channel. Therefore, crosstalk between the A and C signals allows a signal from one color image to appear in another color image. In some embodiments, G and T crosstalk. In some embodiments, the amount of crosstalk between channels is asymmetric. It should be understood that the amount of crosstalk between channels can be controlled (among other things) by selecting signal molecules with appropriate emission spectra and by selecting the size and wavelength range of the detection channels.
[0258] As used herein, “register”, “registering”, and similar terms refer to any process of associating a signal from an image or dataset at a first point in time or viewpoint with a signal from an image or dataset at another point in time or viewpoint. For example, registration can be used to align signals from a set of images to form a template. Similarly, registration can be used to align signals from other images with a template. One signal can be registered directly or indirectly to another signal. For example, a signal from image “S” can be directly registered to image “G”. Similarly, a signal from image “N” can be directly registered to image “G”, or alternatively, a signal from image “N” can be registered to image “S” that has previously been registered to image “G”. Thus, a signal from image “N” is indirectly registered to image “G”.
[0259] As used herein, the term "reference" is intended to refer to a distinguishable point within or on an object. A reference point can be, for example, a marker, a secondary object, a shape, an edge, a region, an irregularity, a channel, a pit, a column, etc. The reference point may exist in an image of the object or in another dataset derived from the detected object. The reference point may be specified by x and / or y coordinates in the object plane. Alternatively or otherwise, the reference point may be specified by z coordinates orthogonal to the xy plane, for example, defined by the relative position of the object and the detector. One or more coordinates of the reference point may be specified relative to the object or image or one or more other analyses in another dataset derived from the object.
[0260] As used herein, the term "optical signal" is intended to include, for example, fluorescence, emission, scattering, or absorption signals. Optical signals can be detected in the ultraviolet (UV) range (approximately 200 nm to 390 nm), visible (VIS) range (approximately 391 nm to 770 nm), infrared (IR) range (approximately 0.771 μm to 25 μm), or other ranges of the electromagnetic spectrum. Optical signals can be detected by excluding all or part of one or more of these ranges.
[0261] As used herein, the term "signal level" is intended to represent the amount or quantity of detected energy or coded information having desired or predefined characteristics. For example, an optical signal can be quantized by one or more of intensity, wavelength, energy, frequency, power, brightness, etc. Other signals can be quantized based on characteristics such as voltage, current, electric field strength, magnetic field strength, frequency, power, temperature, etc. Signal absence is understood as a signal level of zero or a signal level that is not significantly different from noise.
[0262] As used herein, the term "simulation" is intended to represent a representation or model that creates a physical thing or action that predicts the characteristics of that thing or action. In many cases, the representation or model can be distinguished from the thing or action. For example, a representation or model can be distinguished from the thing in terms of one or more characteristics such as color, signal strength detected from all or part of the thing, size, or shape. In a particular embodiment, the representation or model may be idealized, magnified, darkened, or incomplete when compared to the thing or action. Thus, in some embodiments, for example, in terms of at least one of the characteristics described above, the representation of the model can be distinguished from the thing or action it represents. The representation or model may be provided in a computer-readable format or medium, such as one or more of those set forth elsewhere herein.
[0263] As used herein, the term "specific signal" is intended to refer to detected energy or encoded information that is selectively observed relative to other energy or information, such as background energy or information. For example, a specific signal may be an optical signal detected at a specific intensity, wavelength, or color; an electrical signal detected at a specific frequency, power, or field strength; or other signals known in the art to be relevant to spectral and analytical detection.
[0264] As used herein, the term "strip" is intended to refer to a rectangular portion of an object. A strip can be a long, thin band that is scanned by relative motion between the object and the detector in a direction parallel to the longest dimension of the band. Generally, the width of the rectangular portion or strip will be constant along its entire length. Multiple strips of an object can be parallel to each other. Multiple strips of an object can be adjacent to each other, overlap each other, connect with each other, or be separated from each other by gap regions.
[0265] As used herein, the term "variance" is intended to represent the difference between an expected value and an observed value, or the difference between two or more observed values. For example, variance can be the difference between an expected value and a measured value. Variance can be represented using statistical functions such as standard deviation, the square of the standard deviation, the coefficient of variation, etc.
[0266] As used herein, the term "xy coordinate" is intended to represent information specifying position, size, shape, and / or orientation in the xy plane. This information can be, for example, numerical coordinates in a Cartesian system. Coordinates can be provided relative to one or both of the x-axis and y-axis, or relative to another position in the xy plane. For example, the coordinates of an object's analyte can specify the position of the analyte relative to the object's reference or other analytes.
[0267] As used herein, the term "xy plane" is intended to refer to a two-dimensional region defined by the linear axes x and y. When used with a reference detector and the object observed by the detector, this region can be further specified as being orthogonal to the observation direction between the detector and the detected object.
[0268] As used herein, the term "z-coordinate" is intended to indicate information specifying the position of a point, line, or region along an axis orthogonal to the xy-plane. In a particular implementation, the z-axis is orthogonal to the region of the object being observed by the detector. For example, the focal direction of an optical system may be specified along the z-axis.
[0269] In some implementations, affine transformations are used to transform the acquired signal data. In some such implementations, template generation leverages the fact that affine transformations between color channels are consistent across runs. Because of this consistency, a set of default offsets can be used when determining the coordinates of the analytes in the specimen. For example, a default offset file may contain relative transformations (shifts, scales, skews) of different channels relative to a single channel (such as channel A). However, in other implementations, the offsets between color channels drift during and / or between runs, making offset-driven template generation difficult. In such implementations, the methods and systems presented herein can utilize offset-free template generation, which will be further described below.
[0270] In some aspects of the above-described embodiments, the system may include a flow cell. In some aspects, the flow cell includes channels or other configurations of blocks, wherein at least some blocks comprise one or more arrays of analytes. In some aspects, the analytes comprise multiple molecules such as nucleic acids. In some aspects, the flow cell is configured to deliver labeled nucleotide bases to the nucleic acid array, thereby extending primers that hybridize with nucleic acids within the analyte to generate a signal corresponding to the analyte containing nucleic acids. In a preferred embodiment, the nucleic acids within the analyte are identical or substantially identical to each other.
[0271] In some image analysis systems described herein, each image in an image set includes a color signal, where different colors correspond to different nucleotide bases. In some aspects, each image in the image set includes a signal having a single color selected from at least four different colors. In some aspects, each image in the image set includes a signal having a single color selected from four different colors. In some systems described herein, nucleic acids can be sequenced by providing four different labeled nucleotide bases to a molecular array, thereby generating four different images, each containing a signal with a single color, wherein the signal color is different for each of the four different images, thus generating a cycle of four color images corresponding to four possible nucleotides present at specific locations in the nucleic acid. In some aspects, the system includes a flow cell configured to deliver additional labeled nucleotide bases to the molecular array, thereby generating multiple cycles of color images.
[0272] In preferred embodiments, the methods provided herein may include determining whether the processor is actively acquiring data or whether the processor is in a low-activity state. Acquiring and storing large numbers of high-quality images typically requires significant storage capacity. Furthermore, once acquired and stored, the analysis of the image data can become resource-intensive and may interfere with the processing capabilities of other functions, such as ongoing acquisition and storage of additional image data. Therefore, as used herein, the term "low-activity state" refers to the processor's processing power at a given time. In some embodiments, a low-activity state occurs when the processor is not acquiring and / or storing data. In some embodiments, a low-activity state occurs while some data acquisition and / or storage is being performed, but the additional processing power remains constant, allowing image analysis to occur simultaneously without interfering with other functions.
[0273] As used herein, “identifying a conflict” refers to identifying a situation where multiple processes are competing for resources. In some implementations of this kind, one process is assigned a higher priority than another. In some implementations, a conflict may involve the need to prioritize time allocation, processing power, storage capacity, or any other resources that are being prioritized. Therefore, in some implementations, when processing time or capacity needs to be distributed between two processes (such as analyzing a dataset and acquiring and / or storing a dataset), a conflict exists between the two processes, and this conflict can be resolved by assigning priority to one of the processes.
[0274] This document also provides a system for performing image analysis. The system may include a processor; storage capacity; and a program for image analysis, the program including instructions for processing a first dataset for storage and a second dataset for analysis, wherein the processing includes acquiring and / or storing the first dataset on a storage device, and analyzing the second dataset when the processor is not acquiring the first dataset. In some aspects, the program includes instructions for: identifying at least one instance of a conflict between acquiring and / or storing the first dataset and analyzing the second dataset; and resolving the conflict to favor the acquisition and / or storage of image data, such that the acquisition and / or storage of the first dataset is given priority. In some aspects, the first dataset includes image files obtained from an optical imaging device. In some aspects, the system also includes an optical imaging device. In some aspects, the optical imaging device includes a light source and a detection device.
[0275] As used herein, the term "program" refers to instructions or commands that perform a task or process. The term "program" is used interchangeably with the term "module." In some implementations, a program may be a compilation of various instructions executed under the same command set. In other implementations, a program may refer to a discrete batch or file.
[0276] The following describes some surprising effects of utilizing the methods and systems described herein for performing image analysis. In some sequencing implementations, an important measure of sequencing system utility is its overall efficiency. For example, the amount of mappable data generated daily and the total cost of installing and operating the instrument are important aspects of an economical sequencing solution. To reduce the time required to generate mappable data and improve system efficiency, real-time base detection can be enabled on the instrument computer and can be run concurrently with sequencing chemistry and imaging. This allows much of the data processing and analysis to be completed before the sequencing chemistry process is finished. Additionally, it reduces the storage required for intermediate data and limits the amount of data that needs to be transferred over a network.
[0277] While sequence output has increased, the amount of data transferred per run from the system presented in this paper to the network and auxiliary analysis processing hardware has been significantly reduced. Network load is significantly reduced by converting data on the instrument computer (acquisition computer). Without these onboard, offline data simplification techniques, the image output from a suite of DNA sequencing instruments would cripple most networks.
[0278] The widespread adoption of high-throughput DNA sequencing instruments is partly due to their ease of use, support for a wide range of applications, and suitability for virtually any laboratory environment. The efficient algorithms described in this paper allow critical analytical functions to be added to simple workstations with controllable sequencing instruments. This reduction in computational hardware requirements offers several practical benefits that will become even more significant as sequencing output levels continue to increase. For example, image analysis and base detection are performed on simple towers, minimizing heat generation, laboratory footprint, and power consumption. In contrast, other commercial sequencing technologies have recently scaled their computational infrastructure for primary analyses with up to five times the processing power, resulting in a corresponding increase in heat output and power consumption. Therefore, in some specific implementations, the computational efficiency of the methods and systems presented in this paper enables customers to increase their sequencing throughput while keeping server hardware costs to a minimum.
[0279] Therefore, in some specific implementations, the method and / or system proposed herein act as a state machine, tracking the individual state of each specimen and, when it detects that a specimen is ready to proceed to the next state, it performs appropriate processing and advances the specimen to that state. A more detailed example of how the state machine monitors the file system to determine when a specimen is ready to proceed to the next state, according to a preferred implementation, is shown in Example 1 below.
[0280] In preferred embodiments, the methods and systems provided herein are multithreaded and can work with a configurable number of threads. Therefore, for example in the case of nucleic acid sequencing, the methods and systems provided herein can operate in the background during real-time sequencing runs for real-time analysis, or they can run using pre-existing image datasets for offline analysis. In some preferred embodiments, the methods and systems handle multithreading by assigning each thread a subset of the specimens it is responsible for. This minimizes the possibility of thread contention.
[0281] The methods disclosed herein may include the step of obtaining a target image of an object using a detection device, wherein the image comprises a repeating pattern of analytes on the object. Detection devices capable of high-resolution imaging of surfaces are particularly useful. In a particular embodiment, the detection device will have sufficient resolution to distinguish analytes by the density, spacing, and / or analyte size described herein. Detection devices capable of obtaining images or image data from a surface are particularly useful. Exemplary detectors are those configured to maintain a static relationship between the object and the detector while acquiring a region image. Scanning devices may also be used. For example, devices that acquire sequential region images (e.g., so-called “step-through” detectors) may be used. Devices that continuously scan points or lines on the surface of an object to accumulate data to construct a surface image are also useful. Point scanning detectors may be configured to scan points (i.e., small detection areas) on the object surface via raster motion in the xy plane of the surface. Line scanning detectors may be configured to scan lines along the y-dimensional of the object surface, the longest dimension of which occurs along the x-dimensional. It should be understood that the detection device, the object, or both may be movable to achieve scanning detection. Detection devices particularly suitable for applications such as nucleic acid sequencing are described in the following patents: U.S. Patent Application Publications Nos. 2012 / 0270305 A1, 2013 / 0023422 A1, and 2013 / 0260372 A1; and U.S. Patents Nos. 5,528,050, 5,719,391, 8,158,926, and 8,241,573, each of which is incorporated herein by reference.
[0282] The specific embodiments disclosed herein can be implemented as methods, apparatus, systems, or articles of art for producing software, firmware, hardware, or any combination thereof using programming or engineering techniques. As used herein, the term "article of art" refers to code or logic implemented in hardware or computer-readable media such as optical storage devices and volatile or non-volatile memory devices. Such hardware may include, but is not limited to, field-programmable gate arrays (FPGAs), coarse-grained reconfigurable architectures (CGRAs), application-specific integrated circuits (ASICs), complex programmable logic devices (CPLDs), programmable logic arrays (PLAs), microprocessors, or other similar processing devices. In certain specific embodiments, the information or algorithms described herein reside in a non-transitory storage medium.
[0283] In specific implementations, the computer-implemented methods described herein can occur in real time while acquiring multiple images of an object. Such real-time analysis is particularly useful for nucleic acid sequencing applications where nucleic acid arrays undergo repeated cycles of fluid and detection steps. The analysis of sequencing data can often be computationally intensive, making it advantageous to execute the methods described herein in real time or in the background while other data acquisition or analysis algorithms are being performed. Exemplary real-time analysis methods that can be used with the methods of this invention are those applicable to MiSeq and HiSeq sequencing devices commercially available from Illumina (San Diego, Calif.) and / or described in U.S. Patent Application Publication 2012 / 0020537 A1, which is incorporated herein by reference.
[0284] An exemplary data analysis system comprised of one or more programmed computers, wherein the programming is stored on one or more machine-readable media, and the code is executed to perform one or more steps of the methods described herein. In one embodiment, for example, the system includes an interface designed to allow the system to network to one or more detection systems (e.g., optical imaging systems) configured to acquire data from a target object. The interface can receive and process data where appropriate. In a particular embodiment, the detection system outputs digital image data, e.g., image data representing individual picture elements or pixels that together form an image of an array or other object. A processor processes the received detection data according to one or more routines defined by processing code. The processing code can be stored in various types of memory circuitry.
[0285] According to the specific implementation currently envisioned, the processing code performed on the detection data includes a data analysis routine designed to analyze the detection data to determine the location and metadata of individual analytes visible or encoded in the data, as well as locations where no analytes were detected (i.e., locations where no analytes are present, or locations where no meaningful signal was detected from existing analytes). In a particular implementation, analyte locations in the array will generally appear brighter than non-analyte locations due to the presence of fluorescent dyes attached to the imaging analytes. It should be understood that, for example, when the target at the analyte is not present in the array being detected, the analyte does not need to appear brighter than its surrounding area. The color of an individual analyte may depend on the dye used and the wavelength of light used by the imaging system for imaging purposes. Analytes to which the target does not bind or which otherwise lack specific marking can be identified based on other characteristics, such as their expected location in the microarray.
[0286] Once the data analysis routine has located individual analytes in the data, value assignment can be performed. Generally, value assignment will assign a numerical value to each analyte based on the characteristics of the data represented by detector components (e.g., pixels) at the corresponding locations. That is, for example, when processing imaging data, the value assignment routine can be designed to identify the detection of light of a specific color or wavelength at a particular location, as indicated by a group or cluster of pixels at that location. For example, in a typical DNA imaging application, four common nucleotides would be represented by four independent and distinguishable colors. A value corresponding to that nucleotide can then be assigned to each color.
[0287] As used herein, the terms "module," "system," or "system controller" can include hardware and / or software systems and circuits that operate to perform one or more functions. For example, a module, system, or system controller can include a computer processor, controller, or other logic-based device that performs operations based on instructions stored on a tangible, non-transitory computer-readable storage medium such as computer memory. Alternatively, a module, system, or system controller can include a hardwired device that performs operations based on hardwired logic and circuitry. The modules, systems, or system controllers illustrated in the accompanying drawings can represent hardware and circuitry that operate based on software or hardwired instructions, software that directs the hardware to perform operations, or a combination thereof. A module, system, or system controller can include or represent hardware circuitry or circuitry that includes and / or is connected to one or more processors, such as one or more computer microprocessors.
[0288] As used herein, the terms “software” and “firmware” are used interchangeably and include any computer program stored in memory for execution by a computer, including RAM memory, ROM memory, EPROM memory, EEPROM memory, and non-volatile RAM (NVRAM) memory. The memory types described above are merely examples and therefore do not limit the types of memory that can be used to store computer programs.
[0289] In the field of molecular biology, one method used for nucleic acid sequencing is sequencing-by-synthesis. This technology can be applied to massively parallel sequencing projects. For example, by using automated platforms, thousands of sequencing reactions can be performed simultaneously. Therefore, one specific embodiment of the present invention relates to instruments and methods for acquiring, storing, and analyzing image data generated during nucleic acid sequencing.
[0290] The enormous increase in the amount of data that can be acquired and stored makes simplified image analysis methods even more beneficial. For example, the image analysis methods described in this paper allow designers and end users to effectively utilize existing computer hardware. Therefore, in the face of rapidly increasing data output, this paper presents methods and systems for reducing the computational burden of processing data. For example, in the field of DNA sequencing, output has scaled 15-fold in the last year, and hundreds of billions of bases can now be achieved in a single run of a DNA sequencing device. If computational infrastructure requirements grow proportionally, most researchers will still be unable to perform large-scale genome-wide experiments. Therefore, generating more raw sequence data will increase the need for secondary analysis and data storage, making optimization of data transmission and storage extremely valuable. Some specific implementations of the methods and systems presented in this paper reduce the time, hardware, network, and laboratory infrastructure requirements required to generate usable sequence data.
[0291] This disclosure describes various methods and systems for performing these methods. Some examples of the methods are described as a series of steps. However, it should be understood that specific implementations are not limited to the particular steps and / or the order of steps described herein. Steps may be omitted, steps may be modified, and / or additional steps may be added. Furthermore, the steps described herein may be combined, steps may be performed simultaneously, steps may be performed in parallel, steps may be divided into multiple sub-steps, steps may be performed in different orders, or steps (or a series of steps) may be re-executed iteratively. Moreover, although different methods are described herein, it should be understood that these different methods (or steps of these different methods) may be combined in other specific implementations.
[0292] In some specific implementations, a processing unit, processor, module, or computing system “configured to” perform a task or operation can be understood as being specifically constructed to perform a task or operation (e.g., one or more programs or instructions stored thereon or used in conjunction with it are tailored or intended to perform a task or operation, and / or the arrangement of processing circuitry is tailored or intended to perform a task or operation). For clarity and to avoid doubt, a general-purpose computer (which, if properly programmed, can be “configured” to perform a task or operation) is not “configured” to perform a task or operation unless or until it is specifically programmed or structurally modified to perform a task or operation.
[0293] Furthermore, the operations described herein can be complex enough that they cannot be actually performed by a person of ordinary skill or expertise in the art within a commercially reasonable timeframe. For example, these methods may rely on relatively complex calculations that would make them impossible for a person to complete within a commercially reasonable timeframe.
[0294] Throughout this application, various publications, patents, or patent applications have been cited. The disclosures of these publications are incorporated herein by reference in their entirety in order to provide a more complete description of the prior art relating to this invention.
[0295] The term "includes" is intended to be open-ended in this document, encompassing not only the listed elements but also any additional elements.
[0296] As used herein, when referring to a collection of items, the term "each" is intended to identify an individual item in the collection, but not necessarily every item in the collection. Exceptions may occur if explicitly stated otherwise or if the context otherwise specifies otherwise.
[0297] Although the invention has been described with reference to the examples provided above, it should be understood that various modifications may be made without departing from the invention.
[0298] The modules in this application may be implemented in hardware or software and do not need to be precisely divided into identical blocks as shown in the figures. Some of these modules may also be implemented on different processors or computers, or extended across multiple different processors or computers. Furthermore, it should be understood that some of the modules may be combined, operated synchronously, or operated in a different sequence than that shown in the figures, without affecting the implemented functionality. As used herein, the term "module" may include "submodules," which themselves may be considered as constituting modules. The blocks in the figures assigned as modules may also be considered as flowchart steps in the method.
[0299] As used herein, “identifying” an information item does not necessarily require directly specifying that information item. Information can be “identified” in a field by simply referencing the actual information using one or more layers of indirection, or by identifying one or more distinct information items together sufficient to determine the actual information item. Furthermore, the term “specify” is used herein to mean the same thing as “identify.”
[0300] As used herein, a given signal, event, or value is “dependent” on a preceding signal, an event or value of that preceding signal, and an event or value affected by the given signal, event, or value. If an intermediary processing element, step, or time period exists, the given signal, event, or value may still be “dependent” on the preceding signal, event, or value. If an intermediary processing element or step combines more than one signal, event, or value, the signal output of the processing element or step is considered “dependent” on each of the signal, event, or value inputs. If the given signal, event, or value is the same as the preceding signal, event, or value, this is simply a degeneracy case where the given signal, event, or value is still considered “dependent” or “depends on” or “based on” the preceding signal, event, or value. The “responsiveness” of a given signal, event, or value to another signal, event, or value is defined in a similar manner.
[0301] As used in this article, "parallel" or "synchronous" does not require precise simultaneity. It is sufficient if the evaluation of one of these individuals begins before the evaluation of the other of these individuals is completed.
[0302] Computer System
[0303] Figure 17 The present invention provides a computer system 1700 that can be used to implement the technology disclosed herein. The computer system 1700 includes at least one central processing unit (CPU) 1772 that communicates with a plurality of peripheral devices via a bus subsystem 1755. These peripheral devices may include a storage subsystem 1710, which includes, for example, memory devices and a file storage subsystem 1736, a user interface input device 1738, a user interface output device 1776, and a network interface subsystem 1774. The input and output devices allow users to interact with the computer system 1700. The network interface subsystem 1774 provides an interface to an external network, including interfaces to corresponding interface devices in other computer systems.
[0304] In one implementation, the equalizer base detector 104 is communicatively linked to the storage subsystem 1710 and the user interface input device 1738.
[0305] User interface input devices 1738 may include: keyboards; pointing devices such as mice, trackballs, touchpads, or graphics tablets; scanners; touchscreens integrated into displays; audio input devices such as speech recognition systems and microphones; and other types of input devices. Generally, the term "input device" is intended to encompass all possible types of devices and methods for inputting information into computer system 1700.
[0306] User interface output device 1776 may include a display subsystem, a printer, a fax machine, or a non-visual display (such as an audio output device). The display subsystem may include an LED display, a cathode ray tube (CRT), a flat panel device such as a liquid crystal display (LCD), a projection device, or other mechanisms for producing visible images. The display subsystem may also provide a non-visual display, such as an audio output device. Generally, the term "output device" is intended to encompass all possible types of devices and methods for outputting information from computer system 1700 to a user or to another machine or computer system.
[0307] The storage subsystem 1710 stores the programming and data structures that provide the functionality of some or all of the modules and methods described herein. These software modules are typically executed by the processor 1778.
[0308] Processor 1778 can be a graphics processing unit (GPU), a field-programmable gate array (FPGA), an application-specific integrated circuit (ASIC), and / or a coarse-grained reconfigurable architecture (CGRA). Processor 1778 can be powered by deep learning cloud platforms such as Google Cloud Platform. ™ Xilinx ™ and Cirrascale ™ Managed by [the relevant authority]. Examples of processors in the 1778 series include Google's Tensor Processing Unit (TPU). ™ Rackmount solutions (such as the GX4 Rackmount Series) ™ GX17 Rackmount Series ™ NVIDIA DGX-1 ™ Microsoft's Stratix V FPGA ™ , Graphcore’s Intelligent Processor Unit (IPU) ™ Qualcomm's Snapdragon processors ™ ZerothPlatform ™ NVIDIA's Volta ™ NVIDIA's Drive PX ™ , NVIDIA's JETSON TX1 / TX2 MODULE ™ Intel's Nirvana ™ Movidius VPU ™ ,Fujitsu DPI ™ ARM's DynamicIQ ™ IBM TrueNorth ™ It has Testa V100s ™ Lambda GPU servers, etc.
[0309] The memory subsystem 1722 used in the storage subsystem 1710 may include multiple memories, including a main random access memory (RAM) 1732 for storing instructions and data during program execution and a read-only memory (ROM) 1734 for storing fixed instructions. The file storage subsystem 1736 may provide persistent storage for program files and data files and may include hard disk drives, floppy disk drives, and associated removable media, CD-ROM drives, optical disk drives, or removable media enclosures. Modules implementing the functionality of certain embodiments may be stored by the file storage subsystem 1736 within the storage subsystem 1710, or stored in another machine accessible to the processor.
[0310] Bus subsystem 1755 provides mechanisms for enabling various components and subsystems of computer system 1700 to communicate with each other as intended. Although bus subsystem 1755 is schematically shown as a single bus, alternative implementations of the bus subsystem may use multiple buses.
[0311] The computer system 1700 itself can be of various types, including personal computers, portable computers, workstations, computer terminals, network computers, televisions, mainframes, server clusters, a loosely networked group of widely distributed computers, or any other data processing system or user equipment. Due to the constantly evolving nature of computers and networks, [the following applies]. Figure 17 The description of the computer system 1700 depicted herein is intended only as a specific example for illustrating a preferred embodiment of the invention. Many other configurations of the computer system 1700 are also possible, which have... Figure 17 The computer system depicted in the text has more or fewer components.
[0312] Specific implementation
[0313] The technology disclosed in this invention uses equalization-based image processing techniques to attenuate spatial crosstalk from sensor pixels. The technology disclosed in this invention can be practiced as a system, method, or article of manufacture. One or more features of a specific embodiment can be combined with a basic embodiment. Non-exclusive embodiments are taught to be composable. One or more features of a specific embodiment can be combined with other embodiments. This disclosure periodically reminds the user of these options. The omission of statements repeating these options from some embodiments should not be considered as limiting the combinations taught in the foregoing, and these statements are hereby incorporated by reference in each of the following embodiments.
[0314] In one embodiment, the technology disclosed in this invention proposes a computer-implemented method for attenuating spatial crosstalk from sensor pixels.
[0315] The technique disclosed in this invention solves the spatial crosstalk on sensor pixels in a pixel plane caused by periodically distributed fluorescent samples in a sample plane. Signal cones from the fluorescent sample are optically coupled to a local grid of the sensor pixel through at least one lens. These signal cones overlap and impact the sensor pixel, thereby generating spatial crosstalk.
[0316] The technique disclosed in this invention captures the feature extension of a feature signal cone projected through a lens in at least one sub-pixel lookup table, as well as the contribution of the feature signal cone to the fluorescence detected by the sensor pixel in the local grid of the sensor pixel. The local grid of the sensor pixel is substantially concentric with the center of the feature signal cone.
[0317] The technique disclosed in this invention interpolates between a set of subpixel lookup tables that represent feature extensions at subpixel resolution, in order to generate an interpolation lookup table based on the center of the target fluorescent sample.
[0318] The technique disclosed in this invention separates the signal from the target fluorescent sample by convolving an interpolation lookup table with the sensor pixels in the target local grid of the sensor pixels. The signal projects the center of the signal cone approximately onto the center of the target local grid.
[0319] The technique disclosed in this invention uses the sum of the convolutional contributions of the separated signals as the fluorescence intensity from the target fluorescent sample.
[0320] The technique disclosed in this invention then uses fluorescence intensity to detect bases in the first target fluorescent sample. The fluorescence intensity of the first target fluorescent sample is determined for each of the multiple imaging channels. Consider a four-channel chemistry, which uses four imaging channels to generate four images in each sequencing cycle. Then, for the first target fluorescent sample, as described above, four fluorescence intensities are determined using the technique disclosed in this invention. These four fluorescence intensities are then processed by a base detector to detect bases in the first target fluorescent sample. Similarly, for a two-channel chemistry, two fluorescence intensities are used to detect bases in the first target fluorescent sample.
[0321] The methods described in this and other parts of the disclosed technology may include one or more of the following features and / or a combination of the features described in the additional methods disclosed. For the sake of brevity, the combinations of features disclosed in this application are not listed separately and are not repeated with each basic feature group. The reader will understand how easily the features identified in the method can be combined with the basic feature sets identified as implementations in other sections of this application.
[0322] In some embodiments, the periodically distributed fluorescent samples are arranged in a rhomboid shape. In other embodiments, the periodically distributed fluorescent samples are arranged in a hexagonal shape.
[0323] Other specific embodiments of the methods described in this section may include a non-transitory computer-readable storage medium storing instructions executable by a processor to perform any of the methods described above. Yet another specific embodiment of the methods described in this section may include a system comprising a memory and one or more processors operable to execute instructions stored in the memory to perform any of the methods described above.
[0324] In another embodiment, the technology disclosed in this invention proposes a computer-implemented method for base detection.
[0325] The technique disclosed in this invention accesses an image whose pixels depict intensity emissions from a target cluster and intensity emissions from adjacent clusters. These pixels include a center pixel that contains the center of the target cluster. Each of these pixels can be divided into multiple sub-pixels.
[0326] Based on a specific sub-pixel, among a plurality of sub-pixels including the center pixel of the target cluster, the technique disclosed in this invention selects a sub-pixel lookup table corresponding to that specific sub-pixel from a sub-pixel lookup table library. The selected sub-pixel lookup table contains pixel coefficients configured to accept intensity emission from the target cluster and reject intensity emission from neighboring clusters.
[0327] The technique disclosed in this invention multiplies the pixel coefficients element-wise by the intensity values of the pixels in the image, and sums the products of the multiplications to produce an output.
[0328] The technology disclosed in this invention uses this output to detect bases in target clusters.
[0329] Each feature discussed in this particular implementation section for other implementations also applies to this method implementation. As shown above, all method features are not repeated here and should be considered as repeated by reference.
[0330] In some embodiments, the techniques disclosed in this invention further include: (i) selecting an additional subpixel lookup table from a subpixel lookup table library, which corresponds to the subpixel most continuously adjacent to a particular subpixel; (ii) interpolating between the pixel coefficients of the selected subpixel lookup table and the pixel coefficients of the selected additional subpixel lookup table, and generating interpolated pixel coefficients configured to accept intensity emission from the target cluster and reject intensity emission from neighboring clusters; (iii) multiplying the interpolated pixel coefficients element-wise by the intensity values of pixels in the image, and summing the products of the multiplications to produce an output; and (iv) using the output to perform base detection on the target cluster.
[0331] In some embodiments, the target cluster and additional adjacent clusters are periodically distributed in a rhomboid shape on the flow cell and fixed to the orifices of the flow cell. In other embodiments, the target cluster and additional adjacent clusters are periodically distributed in a hexagonal shape on the flow cell and fixed to the orifices of the flow cell.
[0332] In some implementations, the interpolation is based on at least one of linear interpolation, bilinear interpolation, and bicubic interpolation.
[0333] In some implementations, the pixel coefficients of the subpixel lookup tables in the subpixel lookup table library are learned as a result of training the equalizer using a decision-oriented equalization method. In one implementation, the decision-oriented equalization uses least squares estimation as the loss function. In one implementation, the least squares estimation uses ground truth base detection to minimize the squared error. In one implementation, the ground truth base detection is modified to account for DC offset, amplification factor, and multiclonalness.
[0334] In some implementations, the pixel coefficients of the subpixel lookup tables in the subpixel lookup table library are derived from a combination of: (i) a single subpixel lookup table whose pixel coefficients are learned as a result of training an equalizer using a decision-oriented equalization method, and (ii) a pre-computed set of interpolation filters. Each interpolation filter in the set of interpolation filters corresponds to each subpixel among a plurality of subpixels.
[0335] The technique disclosed in this invention also includes making the center of the target cluster substantially concentric with the center of the center pixel by: (i) registering the image relative to a template image and determining affine transformation parameters and nonlinear transformation parameters; (ii) using these parameters to transform the position coordinates of the target cluster and the adjacent clusters to the image coordinates of the image and generating a transformed image with the transformed pixels; and (iii) applying interpolation using the transformed position coordinates of the target cluster and the adjacent clusters so that their respective cluster centers are substantially concentric with the center of the corresponding transformed pixel containing the cluster center.
[0336] The technology disclosed in this invention also includes generating an output for each of a plurality of images captured using a corresponding imaging channel in a specific sequencing cycle, and using the output generated for each image to perform base detection on the target cluster.
[0337] Other specific embodiments of the methods described in this section may include a non-transitory computer-readable storage medium storing instructions executable by a processor to perform any of the methods described above. Yet another specific embodiment of the methods described in this section may include a system comprising a memory and one or more processors operable to execute instructions stored in the memory to perform any of the methods described above.
[0338] This invention discloses the following provisions:
[0339] 1. A computer-implemented method for base detection, the method comprising:
[0340] Accessing its pixels depicts an image of intensity emission from a target cluster and intensity emission from additional neighboring clusters, the pixels including a center pixel containing the center of the target cluster, and each pixel in the pixel can be divided into multiple sub-pixels;
[0341] Based on a specific sub-pixel, among multiple sub-pixels containing the center pixel of the target cluster center, a sub-pixel lookup table corresponding to that specific sub-pixel is selected from the sub-pixel lookup table library. The selected sub-pixel lookup table contains pixel coefficients configured to maximize the signal-to-noise ratio.
[0342] The pixel coefficients are element-wise multiplied by the intensity values of the pixels in the image, and the products of the multiplications are summed to produce an output, where the pixel coefficients act as weights, and the output is a weighted sum of the intensity values; and
[0343] The output is used to detect bases in the target cluster.
[0344] 2. The computer-implemented method according to Clause 1, wherein the signal with maximized signal-to-noise ratio is the intensity emission from the target cluster, and the noise with minimized signal-to-noise ratio is the intensity emission from neighboring clusters.
[0345] 3. The computer-implemented method according to Clause 1, wherein the element-wise multiplication adds a bias to a given set of equalizer coefficients.
[0346] 4. The computer-implemented method according to Clause 3, wherein the deviation is a DC offset as the average value of the background noise intensity.
[0347] 5. The computer-implemented method according to Clause 1 further includes:
[0348] Select an additional subpixel lookup table from the subpixel lookup table library, which corresponds to the most contiguous neighboring subpixel of a particular subpixel;
[0349] Interpolation is performed between the pixel coefficients of the selected subpixel lookup table and the pixel coefficients of the selected additional subpixel lookup table, and interpolated pixel coefficients are generated that are configured to maximize the signal-to-noise ratio.
[0350] The interpolated pixel coefficients are element-wise multiplied by the intensity values of pixels in the image, and the products of these multiplications are summed to produce an output, where the interpolated pixel coefficients act as weights, and the output is a weighted sum of the intensity values; and
[0351] The output is used to detect bases in the target cluster.
[0352] 6. The computer-implemented method according to Clause 1, wherein the target cluster and additional adjacent clusters are periodically distributed in a rhomboid shape on the flow cell and fixed on the orifices of the flow cell.
[0353] 7. The computer-implemented method according to Clause 6, wherein the target cluster and additional adjacent clusters are periodically distributed in a hexagonal shape on the flow cell and fixed to the orifices of the flow cell.
[0354] 8. The computer-implemented method according to Clause 1, wherein the interpolation is based on at least one of linear interpolation, bilinear interpolation, and bicubic interpolation.
[0355] 9. The computer-implemented method according to Clause 1, wherein the pixel coefficients of the subpixel lookup tables in the subpixel lookup table library are learned as a result of training the equalizer using at least one of least squares estimation, ordinary least squares, least average method, and recursive least squares. In other embodiments, other estimation algorithms and adaptive algorithms may be used to train the equalizer.
[0356] 10. The computer-implemented method according to Clause 9 further includes training the equalizer in an offline mode, wherein the pixel coefficients of the subpixel lookup table are fixed after training on multiple batches of training data from previously executed sequencing runs.
[0357] 11. The computer-implemented method according to Clause 10 further includes training the equalizer in an online mode, wherein the pixel coefficients of the subpixel lookup table are iteratively updated as training data from an ongoing sequencing run becomes available.
[0358] 12. The computer-implemented method according to Clause 11 further includes accessing the base-by-base intensity distribution of each of the four bases A, C, G, and T generated during the period prior to base detection of images in training data, selecting the corresponding center of the base-by-base intensity distribution as the base-by-base ground truth target intensity, and using the base-by-base ground truth target intensity to train an equalizer.
[0359] 13. The computer-implemented method according to Clause 12 further includes pre-training the equalizer in offline mode and retraining the equalizer in online mode.
[0360] 14. The computer-implemented method according to Clause 9 further includes generating a lookup table in a subpixel lookup table library by applying a single set of equalizer coefficients and a pre-computed set of interpolation filters together, including interpolating pixel intensities to generate the input to the equalizer. This includes generating the equalizer input by using interpolated pixel intensity values and calculating pixel weights for clusters of alignments that are significantly different from the trained equalizer coefficients relative to the pixels. To efficiently implement the techniques of the present invention with a single shared LUT, the interpolation and equalizer filter responses can be convolved together. In other embodiments, the interpolation filter computation can be performed directly without binning the subpixels.
[0361] 15. The computer-implemented method according to Clause 1 further includes making the center of the target cluster substantially concentric with the center of the center pixel by:
[0362] The image is registered relative to the template image, and the affine transformation parameters and nonlinear transformation parameters are determined.
[0363] The position coordinates of the target cluster and the additional neighboring clusters are transformed into image coordinates of the image using the parameters, and a transformed image with the transformed pixels is generated; and
[0364] Interpolation is applied using the transformed position coordinates of the target cluster and the additional neighboring clusters so that their respective cluster centers are substantially concentric with the center of the corresponding transformed pixel containing the cluster center.
[0365] 16. The computer-implemented method according to Clause 4 further includes generating an output for each of a plurality of images captured using corresponding imaging channels and / or color channels in a particular sequencing cycle, and using the output generated for each image to perform base detection on the target cluster.
[0366] 17. A computer-implemented method for recovering a subsurface signal from a fluorescent sample positioned in a sample plane, the subsurface signal being recovered from a signal disrupted by a surrounding fluorescent source also located in the sample plane, the method comprising:
[0367] In at least one subpixel lookup table, a set of features of illumination in the image plane is captured by the sensor pixel array based on a sampling method that takes into account disruption from the surrounding fluorescence source. Then, when the center coordinates of the fluorescence sample are located at a position distributed on the center pixel of the sensor array, a set of lookup tables for the set of features of illumination generated by the sensor pixel array is generated, the position being distributed relative to the coordinate center of the center pixel.
[0368] The sensor receives an image of the center coordinates of the fluorescent sample at a location in the center pixel of the sensor pixel array, wherein the image is disrupted by the surrounding fluorescence source, and receives the center coordinates of the fluorescent sample within the center pixel.
[0369] An interpolation table is calculated based on the interpolation between lookup tables in the lookup table set to determine the feature set of illumination generated by the sensor pixel array tailored to the received center coordinates of the fluorescent sample.
[0370] The signal is recovered from the target fluorescent sample by multiplying the interpolation lookup table element-wise with the sensor pixels in the target local grid of the sensor pixels. This signal projects the center of the signal cone approximately onto the center of the target local grid.
[0371] The sum of these element-wise multiplications is used as the intensity of the fluorescence from the target fluorescent sample; and
[0372] The fluorescence intensity was used to detect the bases in the first target fluorescent sample.
[0373] 21. A computer-implemented method for base detection, the method comprising:
[0374] Access its pixels to depict an image of intensity emission from the target cluster and intensity emission from adjacent clusters;
[0375] Select a lookup table containing pixel coefficients configured to maximize the signal-to-noise ratio;
[0376] The pixel coefficients are convolved with the intensity values of the pixels in the image to produce an output; and
[0377] Base detection is performed on the target cluster based on the output.
[0378] 22. The computer-implemented method according to Clause 21, wherein the signal whose signal-to-noise ratio is maximized is the intensity emission from the target cluster, and the noise whose signal-to-noise ratio is minimized is the intensity emission from the neighboring cluster, plus additional noise sources.
[0379] 23. The computer-implemented method according to Clause 21, wherein the pixel includes a center pixel containing the center of the target cluster, and each pixel in the pixel can be divided into multiple sub-pixels.
[0380] 24. The computer-implemented method according to Clause 23, wherein the lookup table is a subpixel lookup table.
[0381] 25. The computer-implemented method according to Clause 24 further includes:
[0382] Based on a specific sub-pixel, among a plurality of sub-pixels of the center pixel containing the center of the target cluster, a sub-pixel lookup table corresponding to the specific sub-pixel is selected from a sub-pixel lookup table library, the selected sub-pixel lookup table containing the pixel coefficient;
[0383] The pixel coefficients are element-wise multiplied by the intensity values of the pixels in the image, and the products of these multiplications are summed to produce the output, where the pixel coefficients act as weights, and the output is a weighted sum of the intensity values; and
[0384] Using the output to perform base detection on the target cluster includes generating the output for each of a plurality of imaging channels, and using the output of each imaging channel to perform base detection on the target cluster.
[0385] 26. The computer-implemented method according to Clause 25, wherein the element-wise multiplication adds a bias to a given set of equalizer coefficients, wherein the bias is a DC offset as the average value of the background noise intensity.
[0386] 27. The computer-implemented method according to Clause 25 further includes:
[0387] Select an additional subpixel lookup table from the subpixel lookup table library, which corresponds to the subpixel that is continuously adjacent to the specific subpixel;
[0388] Interpolated pixel coefficients are generated based on the pixel coefficients of the selected subpixel lookup table and the selected additional subpixel lookup table, and the interpolated pixel coefficients are configured to maximize the signal-to-noise ratio;
[0389] The interpolated pixel coefficients are convolved with the intensity values of the pixels in the image to produce an output; and
[0390] Base detection is performed on the target cluster based on the output.
[0391] 28. The computer-implemented method according to Clause 27 further includes:
[0392] The interpolated pixel coefficients are element-wise multiplied by the intensity value of the pixel in the image, and the products of the multiplications are summed to produce the output, wherein the interpolated pixel coefficients act as weights, and the output is a weighted sum of the intensity values.
[0393] 29. The computer-implemented method according to Clause 21 further includes training an equalizer using at least one of least squares estimation, ordinary least squares, least average method, and recursive least squares to generate the pixel coefficients.
[0394] 30. The computer-implemented method according to Clause 29 further includes training the equalizer in an offline mode, wherein the pixel coefficients of the subpixel lookup table are fixed after training on multiple batches of training data from previously executed sequencing runs.
[0395] 31. The computer-implemented method according to Clause 30 further includes training the equalizer in an online mode, wherein the pixel coefficients of the subpixel lookup table are iteratively updated during an ongoing sequencing run.
[0396] 32. The computer-implemented method according to Clause 31 further includes accessing the base-by-base intensity distribution of each of the four bases A, C, G, and T generated during the period prior to base detection of the images in the training data, selecting the corresponding center of the base-by-base ground truth target intensity as the base-by-base ground truth target intensity of the corresponding color channel, and using the base-by-base ground truth target intensity to train the equalizer.
[0397] 33. The computer-implemented method according to Clause 32 further includes pre-training the equalizer in the offline mode and retraining the equalizer in the online mode.
[0398] 34. The computer-implemented method according to Clause 29 further includes generating the lookup table in the subpixel lookup table library by applying a single set of equalizer coefficients and a pre-computed set of interpolation filters together, including interpolating pixel intensities to generate the input of the equalizer.
[0399] 35. The computer-implemented method according to Clause 21 further includes concentrically aligning the center of the target cluster with the center of the center pixel by:
[0400] The image is registered relative to the template image, and the affine transformation parameters and nonlinear transformation parameters are determined.
[0401] The position coordinates of the target cluster and the additional neighboring clusters are transformed into image coordinates of the image using the parameters, and a transformed image with the transformed pixels is generated; and
[0402] Interpolation is applied using the transformed position coordinates of the target cluster and the additional neighboring clusters so that their respective cluster centers are concentric with the center of the corresponding transformed pixel containing the cluster center.
[0403] 36. A non-transitory computer-readable storage medium having computer program instructions applied thereto perform base detection, the instructions, when executed on a processor, implementing a method comprising:
[0404] Access its pixels to depict an image of intensity emission from the target cluster and intensity emission from adjacent clusters;
[0405] Select a lookup table containing pixel coefficients configured to maximize the signal-to-noise ratio;
[0406] The pixel coefficients are convolved with the intensity values of the pixels in the image to produce an output; and
[0407] Base detection is performed on the target cluster based on the output.
[0408] 37. The non-transitory computer-readable storage medium according to Clause 36, wherein the signal with maximized signal-to-noise ratio is the intensity emission from the target cluster, and the noise with minimized signal-to-noise ratio is the intensity emission from the neighboring cluster, plus additional noise sources.
[0409] 38. The method, implemented according to the non-transitory computer-readable storage medium of Clause 36, further comprises training an equalizer using at least one of least squares estimation, ordinary least squares, least average method, and recursive least squares to generate the pixel coefficients.
[0410] 39. A system comprising a memory and one or more processors coupled thereto, the memory being loaded with computer instructions to perform base detection, the instructions, when executed on the processors, performing a plurality of actions, including:
[0411] Access its pixels to depict an image of intensity emission from the target cluster and intensity emission from adjacent clusters;
[0412] Select a lookup table containing pixel coefficients configured to maximize the signal-to-noise ratio;
[0413] The pixel coefficients are convolved with the intensity values of the pixels in the image to produce an output; and
[0414] Base detection is performed on the target cluster based on the output.
[0415] 40. The system according to Clause 39 further includes performing multiple actions including training an equalizer using at least one of least squares estimation, ordinary least squares, least average method, and recursive least squares to generate the pixel coefficients.
[0416] While the invention has been disclosed with reference to the preferred embodiments and examples described in detail above, it should be understood that these examples are intended to be illustrative and not limiting. Modifications and combinations will readily occur to those skilled in the art, and these modifications and combinations will be within the spirit of the invention and the scope of the following claims.
Claims
1. A system comprising: At least one processor; as well as A non-transitory computer-readable medium comprising instructions that, when executed by the at least one processor, cause the system to: For sequencing cycle reception, pixel images depicting intensity emission from the target cluster and intensity emission from neighboring clusters are obtained; Select a set of coefficients corresponding to the intensity emission from the target cluster and the position of the target cluster; The set of coefficients is adjusted for the target pixel to generate pixel-specific coefficients for the target pixel specific to the target cluster; For the target cluster, a correction signal is determined based on the pixel-specific coefficients specific to the target pixel to reduce spatial crosstalk between neighboring clusters on the target cluster; as well as Based on the correction signal of the target cluster, the base detection of the target cluster is determined for the sequencing cycle.
2. The system of claim 1, further comprising instructions, when executed by the at least one processor, to cause the system to receive a pixel image depicting the overlap of intensity emissions from the target cluster with one or more intensity emissions from the adjacent cluster.
3. The system of claim 1, further comprising instructions, when executed by the at least one processor, to cause the system to receive the pixel image depicting the intensity emission from the target cluster and the intensity emission from the adjacent cluster by receiving a pixel image patch depicting a sample planar region including intensity emission from the target cluster and intensity emission from the adjacent cluster.
4. The system of claim 1, further comprising instructions, when executed by the at least one processor, to cause the system to adjust the set of coefficients for the target pixel to generate the pixel-specific coefficients and the set of pixel-specific coefficients for the adjacent clusters by generating sub-pixel-specific coefficients specific to the target pixel representing a feature signal of the target cluster, and a set of sub-pixel-specific coefficients specific to the adjacent clusters of pixels.
5. The system of claim 1, further comprising instructions, when executed by the at least one processor, to cause the system to determine the correction signal for the target cluster based on the pixel-specific coefficients, a set of pixel-specific coefficients specific to the neighboring clusters, and intensity values of intensity emissions from the target cluster and the neighboring clusters in the following manner: According to a predetermined pixel-specific coefficient table, the interpolation pixel-specific coefficients of the pixel array in the pixel image are determined; and The interpolation pixel-specific coefficient is multiplied by the intensity value corresponding to the pixel array.
6. The system of claim 5, further comprising instructions that, when executed by the at least one processor, cause the system to perform: The interpolated pixel-specific coefficients of the pixel array are determined by interpolating between the predetermined pixel-specific coefficient tables of the pixel array customized for the center coordinates of the target cluster; The interpolation pixel-specific coefficients are multiplied element-wise by the intensity values corresponding to the pixel array, thereby multiplying the interpolation pixel-specific coefficients by the intensity values of the pixel array; and Sum the products of element-wise multiplications to determine one or more adjustment intensity values for the target cluster.
7. The system of claim 6, further comprising instructions, when executed by the at least one processor, causing the system to determine the base detection of the target cluster based on the one or more adjustment intensity values of the target cluster.
8. The system of claim 1, further comprising instructions, when executed by the at least one processor, causing the system to determine the correction signal for the target cluster in the following manner: A first adjustment intensity value of the target cluster in the first imaging channel is determined based on the pixel-specific coefficients specific to the target pixel, the set of pixel-specific coefficients specific to the neighboring cluster, and one or more intensity values of intensity emission from the target cluster and the neighboring cluster. A second adjusted intensity value for the target cluster in the second imaging channel is determined based on the pixel-specific coefficients specific to the target pixel, the set of pixel-specific coefficients specific to the pixel group, and one or more intensity values from the intensity emission of the target cluster and the neighboring clusters. as well as The base detection of the target cluster is determined based on the first adjusted intensity value and the second adjusted intensity value.
9. The system of claim 1, further comprising instructions, when executed by the at least one processor, to cause the system to select the set of coefficients by selecting a lookup table comprising pixel coefficients corresponding to the positions of the target clusters.
10. A non-transitory computer-readable storage medium storing instructions that, when executed by at least one processor, cause a computing system to: For sequencing cycle reception, pixel images depicting intensity emission from the target cluster and intensity emission from neighboring clusters are obtained; Select a set of coefficients corresponding to the intensity emission from the target cluster and the position of the target cluster; The set of coefficients is adjusted for the target pixel to generate pixel-specific coefficients for the target pixel specific to the target cluster; For the target cluster, a correction signal is determined based on the pixel-specific coefficients specific to the target pixel to reduce spatial crosstalk between neighboring clusters on the target cluster; as well as Based on the correction signal of the target cluster, the base detection of the target cluster is determined for the sequencing cycle.