Genome sequencing and detection methods

Abstract

A nucleic acid sequencing method is described. For example, sequence data generated by a sequencing device can be analyzed to scan k-mers of a fixed size n in each read in the sequence data. Perfect matches of the k-mers in the sequence data with reference k-mers are identified. K-mer matching can be used to identify alternative alleles in the sequence data with abnormal distributions associated with contamination or other quality issues, and determine quality metrics in real time.

Description

[Technical Field]

[0001] (CROSS-REFERENCE TO RELATED APPLICATIONS) This application claims priority to and the benefit of U.S. Provisional Patent Application No. 63 / 022,296, filed May 8, 2020, the disclosure of which is incorporated herein by reference.

[0002] The disclosed technology relates generally to nucleic acid characterization, e.g., sequencing techniques. In some embodiments, the disclosed technology includes rapid and accurate methods for virus detection from sequence data based on genome sequencing, e.g., whole genome sequencing. [Background technology]

[0003] The subject matter discussed in this section should not be assumed to be prior art merely as a result of its mention in this section. Similarly, it should not be assumed that the problems mentioned in this section, or problems associated with the subject matter provided as background, have been previously recognized in the prior art. The subject matter in this section merely represents different approaches, which themselves may also correspond to embodiments of the claimed technology.

[0004] Next-generation sequencing technology has made sequencing increasingly faster and enabled deeper sequencing depth.However, sequencing accuracy and sensitivity are affected by errors and noise from various sources, such as sample defects or PCR bias during library preparation.Therefore, the detection of very low frequency sequences, such as host samples containing low concentrations of viral or bacterial nucleic acids, can be complicated.Therefore, it is desired to develop a method for detecting and / or sequencing the nucleic acid molecules that exist in small amounts in a fast and accurate manner. Summary of the Invention [Means for solving the problem]

[0005] In one embodiment, the present disclosure relates to a real-time quality control method, the method including: generating sequence data from a biological sample using a sequencing device that performs a sequencing run; identifying k-mers in the sequence data that have perfect matches in a hash table that is initialized with a set of k-mers that includes a reference allele k-mer and an alternative allele k-mer of the reference allele; determining a distribution of the reference allele and the alternative allele in the sequence data based on the number of perfect matches; and generating a quality metric for the biological sample based on the distribution during the sequencing run of the biological sample.

[0006] In another embodiment, the present disclosure relates to a sequencing device including a substrate loaded with a sequencing library prepared from a sample. The sequencing device also includes a computer programmed to: perform a sequencing run on the sequencing device to generate sequence data from the sequencing library; identify k-mers in the sequence data that have perfect matches in a hash table initialized with a set of k-mers including a reference allele k-mer and an alternative allele k-mer of the reference allele; determine the distribution of the reference allele and the alternative allele in the sequence data based on the number of perfect matches; and generate a quality metric for the sequencing device for the biological sample during the sequencing run based on the distribution.

[0007] In another embodiment, the present disclosure relates to a method for detecting variants in a biological sample, the method including generating amplicons from the biological sample using primer pairs, preparing a sequencing library from the generated amplicons, generating sequence data from the sequencing library, identifying sequence reads in the sequence data that initiate within the primer region of a primer of each primer pair and are in the same direction as the primer, trimming the identified sequence reads that are in the same direction as the primer to exclude sequences within the primer region, and identifying variant sequences in the untrimmed sequence reads that span the primer region or are in a different direction from the primer, and at locations in the untrimmed sequence reads that correspond to or are complementary to the primer region.

[0008] The foregoing description is presented to enable one to make and use the disclosed technology. Various modifications to the disclosed embodiments will be apparent, and the general principles defined herein may be applied to other embodiments and applications without departing from the spirit and scope of the disclosed technology. Thus, the disclosed technology is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed herein. The scope of the disclosed technology is defined by the appended claims.

[0009] These and other features, aspects, and advantages of the present invention will be better understood from the following detailed description when read in conjunction with the accompanying drawings, in which like features represent like parts throughout. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is a schematic diagram of a workflow for k-mer alignment, according to an embodiment of the present disclosure. [Figure 2]FIG. 1 is a schematic diagram of an exemplary k-mer of a genome, according to an embodiment of the present disclosure. [Figure 3] FIG. 1 is a schematic diagram of a method for virus detection from sequencing data, according to an aspect of the present disclosure. [Figure 4] FIG. 1 is a schematic diagram of a method for alignment-based virus detection according to aspects of the present disclosure. [Figure 5] FIG. 1 is a schematic diagram of target region or k-mer coverage in alignment-based virus detection, according to an embodiment of the present disclosure. [Figure 6] FIG. 1 is a schematic diagram of a method for generating a set of pathogen-specific k-mers and control k-mers for pathogen detection according to an embodiment of the present disclosure. [Figure 7] FIG. 1 is a block diagram of a system configured to acquire sequencing data and perform alignment-based detection, according to aspects of the present disclosure. [Figure 8] FIG. 1 illustrates an exemplary workflow for sample preparation for pathogen detection. [Figure 9] FIG. 9 shows the sequencing results of the amplicons of the workflow of FIG. 8. [Figure 10] FIG. 1 shows the identification of variants after primer trimming. DETAILED DESCRIPTION OF THE INVENTION

[0011] The following discussion is presented to enable any person skilled in the art to make and use the disclosed technology and is provided in the context of a particular application and its requirements. Various modifications to the disclosed embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other embodiments and applications without departing from the spirit and scope of the disclosed technology. Thus, the disclosed technology is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed herein.

[0012] Various methods and configurations are described herein that enable nucleic acid characterization. In embodiments, the disclosed techniques are used as part of sequence analysis of sequence data generated from biological samples to rapidly and accurately detect genome sequences of interest. In embodiments, the disclosed techniques use an ultrafast hash-based aligner to generate error-reduced or error-free subsequences from the sequence data. One application of the disclosed techniques is the rapid detection of viral genomes present in a sequenced library. The technique operates by scanning each k-mer of a fixed size "n" in all sequence reads of the sequenced library and examining its presence / absence in a hash table. The hash table is initialized with all n k-mers of the viral genome or a curated subset thereof. For example, curation can be used to remove k-mers that are not specific to the pathogen(s) of interest. Successful matches of a sequence k-mer against the hash table are counted for each viral k-mer.

[0013] In embodiments, a specialized aligner employs rapid and complete k-mer matching of virus-specific complete k-mer sets or reduced (e.g., curated) k-mer sets to detect pathogen infections with human positive control amplicons. However, the disclosed techniques may also be used in other applications, such as detecting germline variants in biological samples, characterizing the microbiome, and detecting pooled or complex input samples in environmental monitoring (e.g., wastewater monitoring). Additionally, the disclosed techniques may be used to detect a single pathogen of interest (e.g., SARS-CoV-2) or more than one pathogen in a pathogen panel, such as a respiratory pathogen panel (e.g., SARS-CoV-2, RSV, pneumonia, influenza), or a strain tracking panel containing k-mers representing different strains of a particular pathogen.

[0014] 1 is an exemplary workflow 12 including steps of sample processing by sequence analysis that may be used in conjunction with the disclosed techniques. A sample 20 undergoes processing or sample preparation 24 to generate a sequencing library containing a plurality of nucleic acid fragments suitable for a sequencing step 28 to generate sequence data 30. The sequence data 30 may undergo certain primary analysis steps, such as quality or filtering, before being sent for k-mer scanning and k-mer alignment, as generally provided herein.

[0015] The generated sequence data 30 is scanned to identify k-mers of a fixed size n, and these identified k-mers are provided to a k-mer aligner 36. The k-mer aligner 36 may include a hash table that is initialized with a set 34 of known k-mers of size n derived from a reference genome. The reference genome may be all size n k-mers (or a curated subset thereof) of interest in a pathogen genome or other sequence of interest as provided herein.

[0016] The sequence data 30 may be streamed to the k-mer aligner 36 in real time or on a rolling basis, such that as the sequence data 30 is received, the k-mer aligner 36 operates on additional available sequence data 30 in block 40 to detect k-mers of interest in the sequence data 30. The k-mer aligner 36 identifies k-mers in the sequence data 30 that are perfect matches to the set 34 of k-mers of interest. Perfect matches may be related to the total number of matches for the sample 20. If a sample 20 exceeds a threshold number of perfect matches for the identified k-mers, the workflow 12 provides a detection output 42. In embodiments, individual samples 20 may be characterized as positive or negative for detection of a sequence in the set 34. Because the k-mer aligner 36 operates on the streaming data in real time, the detection function can use perfect matches for k-mers as soon as the threshold number is exceeded to quickly identify the status of the sample 20. Furthermore, k-mer-based detection is less computationally intensive than traditional alignment-based approaches, and in embodiments, less computationally intensive than other k-mer approaches. In one example, the disclosed approach uses a fixed k-mer size n. Thus, k-mer matching is based on matching only k-mers of size n, rather than all k-mers of all possible sizes, or within a range of k-mer sizes. In another example, within the set of all possible k-mers of a fixed size n, the approach evaluates matching against only a known subset based on the known sequence of the reference genome.

[0017] The k-mer counts obtained for each sample 20 are used as provided herein to characterize the sample and provide a detection output 42, e.g., to determine a pathogen infection status. For example, a k-mer count above a threshold indicates a positive result for the presence of a pathogen in the sample. A negative result indicates a k-mer count not at or below the threshold level in the sample. The k-mer counts may be evaluated against a global threshold reflecting the total number of k-mer matches per sample 20. In other embodiments, as disclosed herein, the k-mer counts may be evaluated on a target region basis and / or subjected to quality metrics before correlating the k-mer count and pathogen detection, e.g., a positive or negative result.

[0018] The detection output 42, in embodiments, may include providing a notification, message, or report indicating characteristics of the sample 20, such as a positive or negative detection result. The detection output 42, in embodiments, may control subsequent processing steps of the sequence data 30. In contrast to traditional alignment-based detection, which sends all or most of the input data to secondary analysis, the workflow 12 may limit further processing to the subset of samples that are positive for the pathogen or other genomic / target sequence. That is, once identified, only positive samples 20 may be sent for additional or secondary sequence analysis. In this way, the workflow 12 improves the allocation of processing resources by not directing resources to secondary analysis of samples that may not contain the target sequence based on k-mer matching. The additional sequence analysis may include determining the sequence of the biological sample in block 46 to generate a variant calling output 48. Thus, potentially time-consuming analysis, i.e., alignment and variant calling against a reference genome, may be limited to positive (e.g., infected) samples after identification. Furthermore, samples 20 not yet identified as positive can continue to be evaluated by the k-mer aligner 36 until sufficient data is acquired to confirm a negative or positive result. A further advantage of the disclosed approach is that k-mer-based detection occurs in real time, based on relatively rapid analysis. Thus, processing efficiency is improved without significant delay in initiating secondary analysis on relevant subsets of positive samples. Furthermore, depending on the analytical implementation, the workflow 12 may end after the detection output 42 without proceeding to subsequent analysis or variant calling in block 46.

[0019] 2 is a schematic diagram of k-mers 64 of nucleic acids 60 forming a k-mer set 34 of interest for k-mer aligner 36 (see FIG. 1 ). Nucleic acids 60 can represent all or part of a reference genome or a pre-characterized genome of interest, e.g., a pathogen genome. Thus, the disclosed approach can be reference-free in the sense that a reference genome need not be sequenced along with sample 20; set 34 can be computationally constructed based on stored or accessed reference sequence data for nucleic acids 60. In embodiments, nucleic acids 60 can be reverse complements and / or cDNA copies of a single-stranded reference genome.

[0020] As provided herein, a k-mer or multiple k-mers refers to a contiguous substring or substrings of length "k" contained within a biological sequence, such as a nucleic acid sequence. A set of k-mers can refer to all or only a portion of the sequences contained within a nucleic acid of length L. A known or identified sequence of length L will have all k-mers, and an unidentified or unknown sequence may have xk possible or potential k-mers, where x is the number of possible monomers (e.g., four for DNA or RNA).

[0021] In an embodiment, k-mers are used with a fixed size n, so that for a given operation, all k-mers used to construct the k-mer set 34 and to scan the sequence data are the same fixed size relative to each other. However, different k-mers of the same size represent different sequence strings in different or shifted locations relative to each other. In a particular embodiment, k-mers with length=32 (which can be efficiently analyzed on a 64-bit CPU) are used for k-mer matching, although k-mers of any size with a fixed length greater than 24 may be used. Thus, the fixed length of the k-mers may be 25, 26, 27, 28, 29, 30, etc.

[0022] The nucleic acids 60 may include previously identified sequences, but may also include additional sequences, such as known or predicted variants 70. The disclosed non-referenced approaches take advantage of the fact that variants in viral genomes are very rare relative to the overall size of the virus. During k-mer alignment, k-mers from the sample's sequence data that contain / overlap with a variant are considered "missing" because they cannot have a perfect match in a hash table initialized with a variant-free set 34 of reference k-mers. However, because variants are very rare relative to the overall size of the virus, this simply minimizes the loss of sensitivity. In some methods, known variants present in the population may also be included as one or more "variant k-mers" 34 added to the set 34 of k-mers in the k-mer aligner 36.

[0023] FIG. 3 illustrates an exemplary method 100 for detecting a viral pathogen in a human sample. In the illustrated embodiment, sequence data 102 of the human sample is provided as FASTQ-formatted data, which allows for secondary analysis and alignment of the sequence reads, for example, using DRAGEN or another secondary analysis tool. Alignment 104 of the sequence reads is performed using a k-mer aligner 36 (see FIG. 1 ) to identify k-mer perfect matches of a fixed size n based on the genome of the viral pathogen using a set of reference k-mers. Alignment 104 may also include identifying k-mer perfect matches in the sequence data 102 for one or more human control amplicons (e.g., 2-15 amplicons) used as a measure of sample quality. In some embodiments, alignment 104 may be a conventional DRAGEN alignment against a reference genome including a virus, e.g., SARS-CoV2, and one or more human control amplicons.

[0024] The human reads 110 and viral reads 112 are subjected to additional metrics as provided herein to assess sample quality based on human amplicon coverage 114 and generate a control detection output 120. The metrics also include a viral amplicon coverage metric 130 to provide a viral detection output 132. Positive samples based on both the viral detection output and the control detection output 120 can be sent to variant calling 124 to generate a viral sequence output 128.

[0025] Once alignment / matching of sequence reads using a k-mer aligner 36 is performed, metrics associated with the specified virus are interpreted and judged against detection of virus and internal (human) controls, as shown in Figure 4. In some methods, the number of unique reads 160 mapping to the target region (or detected k-mer) of each amplicon can be counted.

[0026] As shown in FIG. 5 , in embodiments, a "target region" can be defined as the amplicon sequence 184, excluding primers and excluding any overlap with another amplicon 184. This can be done either by a) aligning the reads to the viral genome 180 and counting the number of reads 188 (possibly excluding overlaps) that map to the location of each amplicon, or b) counting the number of k-mers 190 from each amplicon sequence 184 observed in the reads. The number of k-mers or reads is compared to a threshold per amplicon coverage, and each amplicon 184 is referred to as "covered" or "uncovered." If more viral amplicons 184 are covered than a second set threshold, the call or virus detection output is that the virus is detected. The number of all amplicons depends on the assay used. In the example of FIG. 5 , the amplicons 184 are non-overlapping. However, it should be understood that more overlapping amplicons 184 can be used to achieve coverage of the entire viral genome.

[0027] Returning to FIG. 4 , after alignment and / or k-mer identification for the human amplicons 162 and viral amplicons 164, the coverage of each individual human amplicon 170 and coverage of each viral amplicon 172 are counted. The number of reads per amplicon (or number of detected k-mers) is compared to a "target threshold" to determine the covered amplicons. The number of covered amplicons is then used to detect the virus 178 (by a positive detection result based on covered amplicons equal to or greater than the virus threshold) and the internal (human) control 174 (by a positive control detection result based on covered amplicons equal to or greater than the human control threshold). The threshold for detecting positive amplicons, as well as the threshold for the number of amplicons required to detect the control and / or virus, can vary. In some embodiments, the detection threshold can be as little as two amplicons or greater, e.g., three, four, or more amplicons. In embodiments, the threshold number of covered amplicons can be at least 1%, at least 10%, or at least 50% of the total number of amplicons. In embodiments, the threshold number of covered amplicons can range from 1 to 5% of the total number of amplicons in the assay. Because detection is designed to provide rapid results on real-time sequence data as additional sequence data is generated from the sample, setting a percentage threshold allows detection to be based on any combination of positive amplicons. Therefore, detection is independent of sample variation in the position of the sequenced cluster or other detection-specific variables that vary from sample to sample.

[0028] Figure 4 shows exemplary virus detection performed on human controls. For human control amplicons 170, control 1 had 25 unique reads and control 3 had 64 unique reads that exceeded the target threshold, and these amplicons were determined to be covered amplicons. In the next step, the two positive amplicons for the human controls were compared to a human control threshold set at 2 or greater, resulting in a control detection threshold exceedance determination 174. Thus, human control detection 174 involved a two-step analysis in which the coverage of individual human amplicons was determined based on the amplicon coverage threshold, and then the number of amplicons that exceeded the coverage threshold was assessed. Similarly, virus detection 178 involved a first step in which the number of unique reads per virus amplicon (e.g., virus 1, virus 2, virus 3, etc.) was counted. Virus 1 had 34 unique reads, virus 2 had 21 unique reads, and virus amplicon 3 had 64 unique reads and were all considered covered amplicons, while virus amplicon 98, which only had one unique read, was not considered a covered amplicon. In the next step, the three covered amplicons were compared to a virus threshold set at 3 or greater to yield a virus detection result.

[0029] The disclosed approach includes quality and control parameters for establishing a set of reference k-mers and / or control k-mers used in a k-mer aligner (e.g., k-mer aligner 36) for k-mer-based alignment. Figure 6 is a schematic diagram of a method 200 for generating a set of pathogen-specific k-mers and control k-mers for pathogen detection. A given pathogen genome includes a set of all potential k-mers of a fixed size n, where n may be greater than 24 bases. However, certain of these k-mers may have perfect matches in a control genome (e.g., a human genome). Potential pathogen k-mers may be run against the control genome in block 204, and certain k-mers may be removed in block 206 to generate a final set of pathogen k-mers in block 208. In one example, k-mers in the potential set that have perfect matches to the control genome are removed. In another example, k-mers that exceed a similarity threshold with the control genome are removed; for example, k-mers that exceed the similarity threshold can include k-mers with 1-3 bases (contiguous or non-contiguous) that differ from the control genome. For example, for a fixed k-mer size of 32, potential k-mers with 31 / 32 or 30 / 32 sequence matches with the control genome are removed, corresponding to potential base calling errors that could result in false positive detection. Thus, the k-mers retained in the final set in block 208 can include k-mers that do not have a perfect match in the control genome and / or k-mers that have sufficient dissimilarity with the control genome (e.g., 1-3 bases differ within the k-mer).

[0030] A set of control k-mers can be selected from the pool of potential k-mers based on a metric in block 210. In assays in which RNA in a human sample is sequenced to detect the presence of an RNA virus, the human sample will also contain human RNA, e.g., mRNA. Thus, the set of human control k-mers can be based on mRNA sequences that are likely to be consistently expressed in the sample's tissue. The set of control k-mers may be selected to be smaller than the reference set, e.g., include fewer amplicons. The potential sets of control k-mers are run against each other, and in embodiments, against a reference genome, in block 214. Control k-mers that are perfect matches or too similar to each other and the control genome (e.g., differing by 1-3 bases but otherwise a perfect match) are removed in step 216 to generate a final set of control k-mers in block 218. In block 220, the final set of pathogen k-mers and the final set of control k-mers are provided to a k-mer aligner.

[0031] 7 is a schematic diagram of a sequencing device 260 that can be used in conjunction with the disclosed embodiments for obtaining sequence data from a sample as provided herein. The sequencing device 260 can perform a sequencing run on the sample to obtain sequence data. The sequencing device 260 can be implemented according to any sequencing method, such as those incorporating the sequencing-by-synthesis methods described in U.S. Patent Publication Nos. 2007 / 0166705, 2006 / 0188901, 2006 / 0240439, 2006 / 0281109, 2005 / 0100900, U.S. Patent No. 7,057,026, WO 05 / 065814, WO 06 / 064199, and WO 07 / 010251, the entire disclosures of which are incorporated herein by reference. Alternatively, sequencing by ligation techniques may be used in sequencing device 260. Such techniques use DNA ligase to incorporate and identify oligonucleotides and are described in U.S. Patent Nos. 6,969,488, 6,172,218, and 6,306,597, the entire disclosures of which are incorporated herein by reference. Some embodiments may utilize nanopore sequencing, whereby sample nucleic acid strands or nucleotides are removed from the sample nucleic acid by an exonuclease and passed through a nanopore. As a sample nucleic acid or nucleotide passes through the nanopore, each base species can be identified by measuring the fluctuations in the electrical conductance of the pore (U.S. Pat. No. 7,001,792; Soni & Meller, Clin. Chem. 53, 1996-2001 (2007); Healy, Nanomed. 2, 459-481 (2007); Cockroft et al. J. Am. Chem. Soc. 130, 818-820 (2008), the entire disclosures of which are incorporated herein by reference). Yet other embodiments involve detection of protons released upon incorporation of a nucleotide into an extension product.For example, sequencing based on detection of released protons can use electrical detectors and related technology commercially available from Ion Torrent (Guilford, CT, a subsidiary of Life Technologies), or the sequencing methods and systems described in U.S. Patent Application Publication Nos. 2009 / 0026082(A1), 2009 / 0127589(A1), 2010 / 0137143(A1), or 2010 / 0282617(A1), the entire disclosures of each of which are incorporated herein by reference. Certain embodiments can utilize methods involving real-time monitoring of DNA polymerase activity. Nucleotide incorporation can be detected via fluorescence resonance energy transfer (FRET) interactions between a fluorophore-bearing polymerase and a γ-phosphate-labeled nucleotide, or using a zero-mode waveguide, as described, for example, in Levene et al. Science 299, 682-686 (2003); Lundquist et al. Opt. Lett. 33, 1026-1028 (2008); and Korlach et al. Proc. Natl. Acad. Sci. USA 105, 1176-1181 (2008), the entire disclosures of which are incorporated herein by reference. Other suitable alternatives include, for example, fluorescent in situ sequencing (FISSEQ) and massively parallel signature sequencing (MPSS). In certain embodiments, sequencing device 260 may be an iSeq manufactured by Illumina (La Jolla, Calif.). In other embodiments, sequencing device 260 may be configured to operate using a CMOS sensor with nanowells fabricated over photodiodes such that DNA deposits are aligned one-to-one with the respective photodiodes.

[0032] In the depicted embodiment, the sequencing device 260 includes a separate sample substrate 262, e.g., a flow cell or sequencing cartridge, and an associated computer 264. However, as noted above, these may be implemented as a single device. In the depicted embodiment, a biological sample may be loaded onto the substrate 262, which is imaged to generate sequence data. For example, reagents that interact with the biological sample fluoresce at specific wavelengths in response to an excitation beam generated by the imaging module 272, thereby returning radiation for imaging. For example, fluorescent moieties may be generated by fluorescently tagged nucleic acids that hybridize to complementary molecules of the moiety or to fluorescently tagged nucleotides incorporated into oligonucleotides using a polymerase. As will be appreciated by those skilled in the art, the wavelengths at which sample dyes are excited and at which they fluoresce will depend on the absorption and emission spectra of the particular dyes. Such returned radiation may propagate through directing optics. This retrobeam may generally be directed to the detection optics of the imaging module 272, which may be a camera or other optical detector.

[0033] The imaging module detection optics may be based on any suitable technology, such as a charged coupled device (CCD) sensor that generates pixelated image data based on photons impinging on locations within the device. However, it will be understood that any of a variety of other detectors may also be used, including, but not limited to, a detector array configured for time delay integration (TDI) operation, a complementary metal oxide semiconductor (CMOS) detector, an avalanche photodiode (APD) detector, a Geiger-mode photon counter, or any other suitable detector. TDI-mode detection can be coupled with line scanning, as described in U.S. Patent No. 7,329,860, which is incorporated herein by reference. Other useful detectors are described, for example, in the references provided previously herein in the context of various nucleic acid sequencing methodologies.

[0034] The imaging module 272 may be under processor control, for example, by a processor 274, and may also include an I / O control 276, an internal bus 278, non-volatile memory 280, RAM 282, and any other memory configuration capable of storing executable instructions, as well as other suitable hardware components, which may be similar to those described with respect to FIG. 7. Additionally, the associated computer 264 may also include a processor 184, an I / O control 286, a communications module 294, and a memory architecture including RAM 288 and non-volatile memory 290, such that the memory architecture is capable of storing executable instructions 292. The hardware components may be coupled by an internal bus 294, which may also be coupled to a display 296. In embodiments in which the sequencing device 260 is implemented as an all-in-one device, certain redundant hardware elements may be eliminated.

[0035] A processor (e.g., processors 274, 284) can be programmed to assign individual sequencing reads to samples based on the associated index sequence or sequences according to the techniques provided herein. In certain embodiments, based on image data acquired by imaging module 272, sequencing device 260 can be configured to generate sequencing data including sequence reads of individual clusters, each associated with a specific location on substrate 270. Each sequence read can be derived from an insert-containing fragment. The sequencing data includes base calls for each base in the sequencing read. Furthermore, for sequential sequencing reads based on the image data, individual reads can be linked to the same location, and therefore the same template strand, via the image data. In this way, index sequencing reads can be associated with sequencing reads of insert sequences before being assigned to the original sample. Processor 274 can also be programmed to perform downstream analysis on the sequences of a particular sample following the assignment of the sequencing reads to samples.

[0036] ). In certain embodiments, the executable instructions 292 cause the processor to perform one of more actions of the methods disclosed herein. The processor (e.g., processors 274, 284) may be highly reconfigurable field-programmable gate array (FPGA) technology. The processor (e.g., processors 274, 284) may be programmed to receive user input for a particular analysis workflow to access a hash table containing an appropriate set of reference k-mers and / or control k-mers stored in memory (e.g., memories 280, 290). In one example, the device 260 receives user input selecting a run or panel of interest, and the k-mer aligner uses the hash table associated with the user input to align streaming sequences to identify perfect matches of k-mers in the sequencing data. The memory can store multiple different sets of k-mers or different initialization hash tables that are specifically selected based on the user input. In embodiments, the selection may also include the selection of a control k-mer. For example, the control k-mer may include a control k-mer of a human, mammal, or other host organism.

[0037] The disclosed techniques can be used to characterize samples, e.g., biological samples. The sample can be derived from any in vivo or in vitro source, including one or more cells, tissues, organs, or organisms, whether living or dead, or any biological or environmental source (e.g., water, air, soil). For example, in some embodiments, the nucleic acid of the sample comprises or consists of eukaryotic and / or prokaryotic dsDNA originating from or derived from humans, animals, plants, fungi (e.g., molds or yeasts), bacteria, viruses, viroids, mycoplasmas, or other microorganisms. In some embodiments, the nucleic acid of the sample comprises or consists of genomic DNA, subgenomic DNA, chromosomal DNA (e.g., from an isolated chromosome or portion of a chromosome, e.g., from one or more genes or loci from a chromosome), mitochondrial DNA, chloroplast DNA, DNA from (or recombinant DNA contained within) a plasmid or other episome, or double-stranded cDNA made by reverse transcribing RNA using an RNA-dependent DNA polymerase or reverse transcriptase to produce first-strand cDNA and then extending a primer annealed to the first-strand cDNA to produce dsDNA. In some embodiments, the nucleic acid of the sample comprises a plurality of dsDNA molecules in or prepared from a nucleic acid molecule (e.g., a plurality of dsDNA molecules in or prepared from cDNA prepared from genomic DNA or RNA in or derived from a biological (e.g., cell, tissue, organ, organism) or environmental (e.g., water, air, soil, saliva, sputum, urine, feces) source). In some embodiments, the nucleic acid of the sample is from an in vitro source. For example, in some embodiments, the nucleic acid of the sample comprises or consists of dsDNA prepared in vitro from single-stranded DNA (ssDNA) or from single- or double-stranded RNA (e.g., using methods well known in the art, such as primer extension using a suitable DNA-dependent and / or RNA-dependent DNA polymerase (reverse transcriptase)).In some embodiments, the nucleic acid of the sample comprises or consists of dsDNA prepared from all or a portion of one or more double-stranded or single-stranded DNA or RNA molecules using any method known in the art, including methods for: amplification of DNA or RNA (e.g., PCR or reverse transcriptase PCR (RT-PCR), a transcription-mediated amplification method, involving amplification of all or a portion of one or more nucleic acid molecules); molecular cloning of all or a portion of one or more nucleic acid molecules in a plasmid, fosmid, BAC, or other vector that is then replicated in a suitable host cell; or capture of one or more nucleic acid molecules by hybridization, such as hybridization to DNA probes on an array or microarray.

[0038] Advantages of the disclosed method include suppressing noise (e.g., cross-contamination), which appears as uniformly scattered reads throughout the viral genome, as opposed to actual signals clustering by amplicon. This method is adaptable to different amplicons with different PCR performance by setting a variable per-amplicon threshold (higher for strongly amplified amplicons). The disclosed method closely corresponds to existing qPCR tests that also report the number of positive amplicons, and therefore is an output result that is easily converted for clinical use. Detection output per sample may be reported, or the detection output may be subjected to downstream quality control.

[0039] In some embodiments, variant calling data for any positive samples may also be reported. In some embodiments, positive samples may be identified, and the approach includes providing a notification or recommendation for treatment based on the diagnosis of the positive sample. In embodiments, based on a diagnosis of pathogen detection or no pathogen detection by the disclosed approach used as a point-of-care detection system, treatment for the detected pathogen is administered to the patient from whom the sample was collected. For example, if the detected pathogen is based on detection of the SARS-CoV-2 genome, treatment for SARS-CoV-2 is administered or a surveillance protocol is initiated. If the SARS-CoV-2 genome is not detected, a SARS-CoV-2 vaccine may be administered based on a diagnosis of no active infection.

[0040] Further advantages of the disclosed approach include real-time quality metrics and variant detection generated on-board. In the example of FIG. 7, the real-time quality metrics are generated on the sequencing device 260 and not as part of a cloud-based secondary analysis. In certain embodiments, sequence data can be analyzed based on the presence and distribution of variants, which may include alternative alleles or single nucleotide polymorphisms (SNPs). The assay may include amplicon generation and / or targeted sequencing based on the desired variant or SNP analysis. For any particular detected allele, the distribution of the allele within the sequence reads can be evaluated to obtain quality metrics on-board. Allele detection can be alignment-based or use k-mer matching as provided herein. The k-mer approach can be extended to detect known variants by including alternative allele versions of the k-mer, where some allele k-mers represent references and others represent alternative alleles. The reference allele k-mer and the alternative allele k-mer can include respective k-mer sets spanning the variant sequence or sequence location. These modifications allow an alignment-free (and therefore faster) version of the algorithm to be run while the sequencing device 260 is still generating data.

[0041] For a given variant and a given individual sample, the distribution of alleles can follow a predictable level. In one example, a particular germline variant allele, if present, is likely to be present in 50% of the sequence reads (50% of sequence reads have one allele, and the remaining 50% have the other allele) or 100% of the reads. Furthermore, if a germline variant is not detected, it is likely to be present in 0% of the sequence reads. Therefore, a variant-to-reference ratio of 1:1 or 1:0 can be considered within the expected distribution of a detected germline variant. However, a distribution of 80% to 20% or 95% to 5% within the reads for an individual sample is biologically unlikely and therefore potentially the result of error or contamination. Therefore, ratios that deviate from a 1:1 or 1:0 ratio (e.g., within a 5% to 10% tolerance range to account for sequencing error) are likely to be a sequencing artifact and / or due to sample contamination. Thus, the sequencing device 260 can assess the sample with respect to germline variant allele distribution based on variant detection in the sequence reads.

[0042] For a given variant panel, e.g., an SNP panel, only some of the variants may be matched to a particular sample. However, for variants detected that deviate from the expected allele distribution, the abnormal distribution may be an indication of sample contamination, a patient or sample lead assignment error, or a sample preparation problem. Thus, samples containing variants with abnormal or low-frequency distributions (e.g., 95% to 5%) may be flagged. In response to flagging, the sequencing device 260 may display an error message (e.g., a displayed notice) on a graphical user interface that identifies potentially contaminated samples in real time. Thus, the disclosed approach includes real-time sample quality metrics for the sequencing device 260. Samples may be indicated as passing or failing depending on one or more assessed allele distributions. In embodiments, simply failing one allele distribution is sufficient to flag the sample. In the case of k-mer-based detection, the k-mer sets generated by the calculation of variants or alternative alleles may be updated as new variants or strains are tracked.

[0043] Identification of a flagged or failed sample based on abnormal allele distribution can cause the sequencing device 260 to stop transmitting the sample's associated sequence data to a cloud-based secondary analysis. Thus, for a multiple sample or multiplexed run, the sequencing device 260 can transmit only passing samples to the cloud for further analysis. If multiple samples all contain the same abnormal allele distribution, the entire multiplexed run can be flagged as potentially contaminated.

[0044] In embodiments, the disclosed techniques include improved detection of variants that may be masked in sequencing data based on primer design or location. For example, variants in regions of the genome corresponding to primer regions may be identified based on overlapping amplicon design, whereby the primer regions are covered by genomic reads from the overlapping amplicons. Figure 8 shows an exemplary workflow for sample preparation for pathogen detection that may be used in conjunction with the disclosed variant detection techniques. In the illustrated example, the sample is processed to extract RNA at block 300. RNA can be extracted from a sample such as a nasopharyngeal swab.

[0045] The extracted RNA is converted to cDNA, and the cDNA is used to generate amplicons using assay-specific primer sets. For example, in COVIDSeq applications, the cDNA is split into two portions, and two different primer pools are used to generate overlapping amplicons 304 that differ between the two portions. Each sample is indexed, e.g., via tagmentation, in step 308, and sequenced in step 310.

[0046] Figure 9 shows the amplicon sequencing results of the workflow in Figure 8, showing overlapping coverage for amplicons 314 and 316 generated from a first and second primer pools for sequence reads that are both indexed, e.g., indexed as originating from the same sample. Pool 1 reads 320 and pool 2 reads 324 contain overlapping portions at their edges within the primer region. Forward reads 326 and reverse reads 328 are present within pool 1 reads 320 and pool 2 reads 324. Post-PCR fragmentation has the effect of partially depleting primers, resulting in an edge effect where reads are clustered toward the primers in both the forward and reverse directions. Sequence reads in the overlapping region contain a heterogeneous mixture of genomic reads with variants and primer reads with reference sequences.

[0047] However, because primer reads represent an expanded portion of the mixture due to edge effects, clustering of primer reads toward the ends of amplicons can reduce the fraction of observed alternative alleles. For example, forward primer 330 overlaps with an internal region of another amplicon 334. All pool 1 reads 320 are forward reads 326 derived from primer 330. Pool 2 reads 324 contain both forward reads 326 and reverse reads 328. Pool 1 reads 320 derived from primer 330 are considered to be perfect matches of the primer and therefore do not contain variants present in the region of the genome covered by primer 330.

[0048] To improve the sensitivity of variant detection, the disclosed technique includes a primer trimming step that hard-clips, masks, or removes primer sequences from reads. The filter trims 1) reads that start in the primer region and 2) reads that match the primer orientation. That is, any sequence read that starts in the region covered by the primer and has a first nucleotide that is either a forward read in the forward primer direction or a reverse read in the reverse primer direction is trimmed. However, coverage in the primer region remains from overlapping amplicons spanning the read and any opposite-strand (complementary) reverse reads. Figure 10 shows an example of trimmed reads in the primer-covered region of a reaction. The read mixture includes untrimmed or retained reads 350 and trimmed reads 352. Reads 352 are trimmed only at the sequence corresponding to primer region 354, indicated by the start X and end X; only forward reads are trimmed. As shown, retained reads 350 contain primarily G-to-T variants. The T variant is not observed in most of the trimmed reads. In one example, variants are called based on a threshold percentage (e.g., at least 50%) of untrimmed reads 350 that contain the variant.

[0049] Table 1 shows an example of improved detection of a single G to T variant. After filter trimming, the remaining allele fractions converged toward nearly 100% allele fraction in the remaining reads, which is considered to be the expected biological distribution.

[0050] [Table 1]

[0051] Although the depicted embodiment shows the trimming of a single primer, primer trimming can be used to cover all primers in a reaction, i.e., both forward and reverse, to improve identification of variants in any region covered by the primers. For whole genome sequencing of pathogens where multiple primer pairs are used, e.g., 50 or more, primer trimming can significantly improve variant detection.

[0052] This written description uses examples, including the best mode, of the disclosed embodiments and enables any person skilled in the art to practice the disclosed embodiments, including making and using any device or system and performing any incorporated methods. The patentable scope of the disclosure is defined by the claims, and may include other examples that occur to those skilled in the art. Such other examples are intended to be within the scope of the claims if they include structural elements that do not differ from the literal words of the claims, or if they include equivalent structural elements that have insubstantial differences from the literal words of the claims. [Explanation of symbols]

[0053] 12 Workflow 20 samples 24 Sample Preparation 28 Sequencing 30 Sequence Data 36 k-mer aligners 42 Detection output 48 Variant Calling Output 60 Nucleic acids 70 known or predicted variants 102 Sequence Data 104 Alignment 110 Human Lead 112 viral reads Coverage of 114 human amplicons 120 Control detection output 124 Variant Calling 128 viral sequence outputs Coverage metrics for 130 viral amplicons 132 Virus detection output 160 unique leads 162 Human Amplicons 164 viral amplicons Coverage of 170 human amplicons Coverage of 172 viral amplicons 174 Human Control Detection 178 Virus Detected 180 viral genomes 260 Sequencing Device 262 Sample Substrate 264 Computer 270 Base material 272 Imaging Module 274 processors 276 I / O control unit 278 Internal Bus 280 Non-volatile Memory 284 processors 286 I / O control unit 290 Non-volatile memory 292 executable instructions 294 Communication Module 296 displays 314 Amplicons generated from the first primer pool 316 Amplicons generated from the second primer pool 320 leads 324 leads 326 Forward Lead 328 Reverse Lead 330 Forward Primer 350 Untrimmed or Retained Leads 352 Trimmed Reeds 354 primer region

Claims

1. 1. A real-time quality control method, comprising: generating sequence data from the biological sample using a sequencing device that performs a sequencing run; identifying k-mers in the sequence data that have perfect matches in a hash table initialized with a set of k-mers including a reference allele k-mer and alternative allele k-mers of the reference allele; determining the distribution of the reference allele and the alternative allele at the identified k-mer; generating a quality metric for the biological sample based on the distribution during the sequencing run of the biological sample; if the quality metric indicates a failing quality metric, stopping communication of the sequence data from the sequencing device; A method comprising:

2. 10. The method of claim 1, comprising flagging the biological sample as contaminated based on the failing quality metric.

3. The method described in claim 2, wherein the biological sample is flagged as contaminated based on the quality metric indicating that the alternative allele is present in less than 5% of sequence reads of the sequence data.

4. The method of claim 1 , wherein the quality metric is generated on the sequencing device.

5. 10. The method of claim 1, wherein the alternative allele comprises a single nucleotide polymorphism.

6. 1. A sequencing device comprising: a substrate loaded with a plurality of sequencing libraries prepared from a biological sample; A computer, causing the sequencing device to perform a sequencing run to generate sequence data from a plurality of sequencing libraries; identifying k-mers in the sequence data that have perfect matches in a hash table initialized with a set of k-mers including a reference allele k-mer and alternative allele k-mers of the reference allele; determining, for each of the biological samples, a distribution of the reference allele and the alternative allele at the identified k-mer; generating quality metrics on the sequencing device for the individual biological samples during the sequencing run based on the distribution, the biological samples comprising a first set of individual biological samples having passing quality metrics and a second set of the individual biological samples having failing quality metrics; and, a computer programmed to perform the steps of: a communication circuit that communicates the generated first set of sequence data to a cloud computing environment and does not communicate the generated second set of sequence data; A sequencing device comprising:

7. The sequencing device of claim 6 , comprising a display that displays the quality metric.

8. The sequencing device of claim 6, wherein the passing quality metric is associated with the ratio of the reference allele to the alternative allele being within an expected range.

9. The sequencing device of claim 6, wherein the failing quality metric is associated with the alternative allele being present in less than 5% of the sequence reads of the sequence data.

Images