PCR system and method for automatic threshold setting

Abstract

Embodiments of digital polymerase chain reaction (dPCR) systems and methods with improved automatic threshold performance and accuracy are disclosed. One embodiment of the present disclosure includes a dPCR processing module that includes an automatic threshold proxy that dynamically analyzes and selects between Gaussian mixture model (GMM) and K-means clustering for automatic threshold results of dPCR assays using a dPCR instrument to improve dPCR measurement techniques.

Description

[0001] Cross-reference to related applications

[0002] This application claims the benefit of U.S. Provisional Application No. 63 / 460,882, filed April 20, 2023. This application is also subject to the following jointly assigned provisional applications: U.S. 63 / 460,884, U.S. 63 / 460,879, and U.S. 63 / 460,885, all filed April 20, 2023. The entire contents of the foregoing applications are incorporated herein by reference. Background Technology

[0003] This disclosure generally relates to image processing systems and methods for detecting (e.g., quantifying) the presence of a target analyte in a sample using digital polymerase chain reaction (dPCR).

[0004] dPCR can be performed on biological samples containing or suspected of containing the target analyte, such as specimens like cells, tissues, or hair; biological fluids such as blood, urine, and saliva; cell clusters such as microbial colonies; or organisms, cells, microorganisms, bacteria, viruses, proteins, antibodies, or nucleic acids such as DNA or RNA molecules. Target analytes include the “original” analyte initially present in the biological sample and any “synthetic” analyte indicating the presence of the original analyte that may be added to or generated during detection, including PCR amplicon, antigen-antibody complexes, etc. Digital quantification (e.g., digital PCR) begins with a sample containing a relatively small amount of the target analyte, such as a polynucleotide or nucleotide sequence template DNA (or RNA). The sample is divided into a large number of smaller test samples, which will ideally contain one (or a small amount close to one) of the target analyte or contain no target analyte, allowing for separate detection reactions in each partition. Appropriate partitioning allows for sufficient distance between individual targets to enable independent detection or quantification. The separating region may or may not include a separating barrier, such as a wall or membrane, or a liquid or semi-solid medium immiscible with the sample. Exemplary separating regions include individually separated targets, such as those deposited on a substrate, glass slides, tubes, open or closed pores, droplets, vesicles, chambers, or beads, or any representation of a separate signal from the target that is distinguishable against background or noise (e.g., a bright spot on a dark background in a digital or analog image). In digital PCR methods, when a sample is thermally cycled using a PCR device, samples containing the target are amplified and produce a positive detection signal, while samples without the target are not amplified and do not produce a detection signal. After multiple PCR amplification cycles, the sample is imaged and subjected to fluorescence analysis to quantify the target concentration in the sample.

[0005] Typically, dPCR involves dispensing a PCR solution containing the sample into tens of thousands of nanoliter droplets, with each droplet undergoing a separate PCR reaction. The PCR solution may include, for example, template DNA (or RNA), a fluorescent quencher probe, primers, and a PCR premix containing DNA polymerase, dNTPs, MgCl2, and an optimal concentration of reaction buffer. Several different methods can be used to dispense the sample, including microplates, capillaries, emulsifiers, and microchamber arrays with nucleic acid-binding surfaces. This dispensing of samples allows for more reliable acquisition and more sensitive measurement of nucleic acid levels. This method has been shown to be useful for studying variations in gene sequences, such as copy number variations and point mutations, and is commonly used for clonal amplification of next-generation sequencing samples.

[0006] Typically, dPCR instruments consist of microplates with a large number of wells (e.g., over 1,000, over 20,000, over 100,000, etc.). Fluorescent dyes act as intercalating agents, emitting detectable fluorescence when they bind to DNA (e.g., human DNA or pathogenic DNA). As PCR progresses and the amount of DNA increases, more dye binds to the PCR products, thus increasing the signal intensity. dPCR instruments also include a thermal cycler for amplification and a light source for exciting fluorescent probes associated with multiple wells. A camera records the fluorescence reaction of multiple wells and transmits the images to a processor-based device that controls the instrument, detects the fluorescence dataset, and processes the data for further analysis. Specifically, after a specified number of PCR amplification cycles (designated as the "endpoint"), the fluorescence of the sample wells is examined, and based on a comparison with a threshold level, the fluorescence value of a given sample well is determined to represent the presence (amplification) or absence (no amplification) of the target analyte. Determining an appropriate threshold level is crucial for obtaining accurate results using dPCR technology. Summary of the Invention

[0007] Given that dPCR assays typically generate thousands or even hundreds of thousands of individual PCR reactions and corresponding fluorescence measurements, efficient computerized processing of the resulting data is crucial for making dPCR a useful technique. Automated computer-based threshold analysis of dPCR data can improve dPCR throughput and accuracy, thereby enhancing the efficiency and performance of dPCR instruments.

[0008] One automated thresholding method relies on automated clustering analysis. Clustering analysis can be used to determine appropriate thresholds to distinguish between two (or more) clusters of data points corresponding to a base group, or to determine whether a specific data point belongs to a single cluster corresponding to a base group.

[0009] Typically, a specific clustering analysis method best suited for a given application is selected. However, embodiments of this disclosure improve dPCR technology by implementing computerized automated thresholding analysis that intelligently selects between different clustering algorithms based on a resolution determination related to the clustering analysis of a given group of dPCR fluorescence measurements. In some embodiments, a threshold formula associated with the results from the selected clustering algorithm is used to set an automated threshold for distinguishing fluorescence values ​​representing amplification and non-amplification in a given partition.

[0010] Gaussian Mixture Models (GMMs) (e.g., the sklearn.mixture Gaussian Mixture Python library) are a useful method for automated clustering analysis applied to dPCR data. A Gaussian Mixture Model is a probabilistic model that assumes all data points are generated by a mixture of a finite number of Gaussian distributions with unknown parameters. If two clusters are assumed, a GMM can fit two Gaussian distributions to the dPCR data, assigning a probability to each point belonging to a given cluster, and can return the mean and standard deviation of each cluster. The mean and standard deviation of each cluster can be used to calculate the threshold that distinguishes the two clusters.

[0011] However, GMMs do not always identify two clusters in dPCR data with sufficiently high resolution. In such cases, it is possible to assume one cluster (instead of two) and assign a probability to each data point, defining the likelihood that the data point belongs to that cluster. This may be an appropriate assumption for some dPCR datasets, for example, if the target analyte is completely absent from the sample, and the amplification results of all sample partitions indicate the absence of the target. Alternatively, the bicluster assumption can be maintained, but in this case, the risk of error in setting the threshold for assigning binary classification using GMMs may be unacceptably high for many data points.

[0012] Embodiments of this disclosure recognize that while GMM provides optimal results in many cases, for some dPCR datasets, other clustering algorithms or mathematical models (e.g., K-means clustering) can better distinguish between two clusters. Therefore, dynamically selecting between two clustering models during automatic thresholding can improve the accuracy of dPCR techniques. With this in mind, some embodiments of this disclosure dynamically select between different clustering models for automatic thresholding analysis of dPCR results. In some embodiments, resolution analysis is used to dynamically determine whether to use GMM clustering or K-means clustering for thresholding.

[0013] It should be understood that embodiments of this disclosure can be implemented in various ways, such as processes, apparatus, systems, devices, articles of manufacture, or methods. Details of several embodiments are further described below. Further details of embodiments of systems and methods consistent with this disclosure are disclosed herein. Brief description of the attached diagram

[0014] This application can be best understood by referring to the embodiments described below in conjunction with the accompanying drawings, in which the same parts are indicated by the same numbers.

[0015] Figure 1A A networked dPCR system according to one embodiment of the present disclosure is shown, which has an improved performance-enhanced automatic threshold agent.

[0016] Figure 1B Figure 1 is a block diagram of server automatic threshold proxy according to some embodiments.

[0017] Figure 2A Figure 1 is a flowchart of the automatic thresholding method of the dPCR operating system according to some embodiments.

[0018] Figure 2B Applications according to some embodiments Figure 2A The flowchart shows the GMM fitting method for site statistics using the automatic thresholding method.

[0019] Figure 2C Applications according to some embodiments Figure 2A The flowchart shows the K-means fitting method for site statistics using the automatic thresholding method.

[0020] Figure 2D Based on the use of some embodiments Figure 2B The flowchart shows the method for calculating a group threshold using the automatic thresholding method of GMM.

[0021] Figure 2E This is a flowchart of a method for calculating two group thresholds using the automatic thresholding method of Figure 2 according to some embodiments.

[0022] Figure 2F This is a flowchart of a method for calculating thresholds for two groups using the K-means of the automatic thresholding method in Figure 2, according to some embodiments.

[0023] Figure 3 A diagram illustrating an example computing device that can implement the examples described herein.

[0024] Detailed description

[0025] Various embodiments will now be described more fully below with reference to the accompanying drawings, which form part of the invention and serve as illustrative examples of practical embodiments. However, this specification may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this specification will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art. In addition, this specification may also be embodied in the form of methods or apparatus. Therefore, any embodiment in the various embodiments herein may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Therefore, the following description should not be considered limiting.

[0026] In some embodiments, the dPCR operating system and the method for processing biosampling-related datasets distributed in a microplate microchamber array include a processor-based automated thresholding (automatic thresholding) agent with improved automated thresholding analysis performance and higher accuracy in setting threshold fluorescent dyes. Specifically, this dPCR operating system with improved automated thresholding performance and accuracy comprises a dPCR processing module that includes an automated thresholding agent using a Gaussian mixture model (GMM) and K-means clustering. The automated thresholding method includes determining whether a cluster of one or two data points exists within the associated fluorescence data corresponding to multiple optical signals detected at the dPCR instrument. In one example, the method includes using GMM to detect the presence of two clusters with two distinct groups. When two clusters are detected, the GMM threshold for both groups is specified as the final threshold. If two distinct groups are not detected using GMM, the method uses K-means clustering to detect the presence of two clusters. When two clusters are detected, a two-group K-means clustering threshold (K-means threshold) is specified as the final threshold. When two clusters are not detected, a single-group GMM threshold is specified as the final threshold.

[0027] In one example, a dPCR operating system with improved automatic thresholding performance and accuracy comprises a server coupled to a dPCR instrument having at least one microplate. The server includes at least one processor coupled to at least one memory and a dPCR processing module containing an automatic thresholding agent. The automatic thresholding agent has a cluster detection unit for the receiver, a Gaussian Mixture Model (GMM) engine, and a K-means clustering engine. The receiver, coupled to at least one memory storing computer-readable instructions, causes the at least one processor, when executed, to receive fluorescence data from the dPCR instrument corresponding to multiple optical signals detected by the dPCR instrument. The fluorescence data includes multiple individual points associated with multiple microchambers, where the multiple individual points form one or more clusters sharing relative characteristics. The GMM engine is coupled to the receiver and at least one memory with instructions that cause the at least one processor to use the GMM to detect whether one or more clusters are two clusters associated with two different groups. For the case where two clusters are detected using the GMM, the GMM engine cooperating with the processor specifies a bi-group GMM threshold as the final threshold. If two distinct groups are not detected using GMM, the processor-cooperative K-means clustering engine can operate to use K-means clustering to detect whether the number of one or more clusters is related to two clusters in a bi-cluster group. If two clusters are detected using K-means clustering, the processor-cooperative K-means clustering engine can operate to specify the bi-cluster K-means clustering threshold as the final threshold. If two clusters are not detected using K-means clustering, the processor-cooperative K-means clustering engine can operate to specify the single-cluster GMM threshold as the final threshold.

[0028] Other aspects and advantages of the embodiments will become apparent from the following detailed description taken in conjunction with the accompanying drawings, which illustrate the principles of the described embodiments by way of example.

[0029] Figure 1AA networked digital polymerase chain reaction (“dPCR”) system 100 according to one embodiment of the present disclosure is shown, featuring an improved automatic threshold agent 130. System 100 includes a dPCR instrument 102, at least one client node 160, and a server 110. The digital PCR instrument 102 includes a microplate (not shown) having an array of microchambers filled with biological samples. Instructions for performing dPCR reside in the automatic threshold agent 130 (computer program product) within a dPCR processing module 120, which is stored in storage device 116, and these instructions are executable by a processor 114. While the processor 114 is executing instructions for the dPCR processing module 120, the instructions, or portions thereof, are typically loaded into working memory 112, from which the processor 114 can easily access the instructions. In the illustrated embodiment, the automatic threshold agent 130 is stored in local storage device 118 or another non-transitory computer-readable medium (which may contain media distributions across different devices and locations).

[0030] Although not shown, the dPCR instrument 102 includes a microplate (not shown) with multiple compartments containing a fluorescent dye solution. The dye acts as an intercalating agent, emitting detectable fluorescence when bound to double-stranded DNA or a pathogen. As PCR proceeds and the amount of double-stranded DNA increases, more dye binds to the PCR product, thus increasing the signal intensity. The digital dPCR instrument 102 also includes a thermal cycler (not shown) for amplification and a light source (not shown) for exciting fluorescent probes associated with the multiple compartments. A camera (not shown) is connected to record the fluorescence reaction of the multiple compartments and transmits the images to a dPCR processing module 120, which controls the instrument, detects the fluorescence dataset, and performs automated threshold analysis. Specifically, after multiple PCR amplification cycles, the fluorescence of the sample is examined, where the dPCR processing module 120 detects binary readings of "0" or "1" associated with the microchamber array.

[0031] In one embodiment, processor 114 may include multiple processors, which may include additional working memory (the additional processors and memory are not shown separately), and a graphics processing unit (GPU) comprising at least several thousand arithmetic logic units supporting massively parallel computing. GPUs are commonly used in deep learning applications because they can perform relevant processing tasks more efficiently than typical general-purpose processors (CPUs). Other embodiments include one or more dedicated processing units comprising systolic arrays and / or other hardware arrangements supporting efficient parallel processing. In some embodiments, such dedicated hardware works in conjunction with a CPU and / or a GPU to perform the various processes described herein. In some embodiments, such dedicated hardware includes application-specific integrated circuits (ASICs, which may refer to a portion of an application-specific integrated circuit), field-programmable gate arrays, and combinations thereof. However, it should be noted that implementation using one or more general-purpose processors (preferably multi-core architectures) such as processor 114 does not depart from the spirit and scope of the invention.

[0032] Client node 160 includes a display 162 for displaying the processing results performed by dPCR processing module 120. In alternative embodiments, an automatic threshold agent 130, or a portion thereof, may be stored in a storage device and executed by one or more processors residing on PCR instrument 102 and / or client node 160. Such alternatives do not depart from the scope of the invention.

[0033] In some embodiments, system 100 includes a network 140 coupled between server 110 and client node 160. Although not shown, in some embodiments, server 110 and client node 160 are directly coupled to each other without network 140. The system may also include at least one dumb terminal (not shown), such as a landline telephone, mobile phone, pager monitor, etc. Although not shown, server 110 may also be coupled to a regular telephone (not shown) via the Public Switched Telephone Network (PSTN) coupled to network 110. Server 110 may be coupled to a remote storage device 150 for referencing versions of previous PCR analysis results and other parameters described below. In some embodiments, each client node 160 includes a dPCR / qPCR agent 104, a data analysis system 102, a memory (not shown), a processor (not shown), and local data storage (not shown).

[0034] In some embodiments, server 110, having dPCR processing module 120 and automatic thresholding agent 130, communicates with each client node 160 and acts as the sole agent for performing the PCR method described herein. In other embodiments, a copy of the automatic thresholding agent (not shown) within client node 160 serves as a device for communicating with server 110 to perform the automatic thresholding method, described in more detail below, in real time. One or more client nodes 160, server 110, and remote storage device 150 may be located in the same local area network (LAN) or in different LANs that can be coupled via the Internet but are separated by firewalls, routers, and / or other network devices. In one embodiment, one or more client nodes 160 are connected to network 110 via a mobile communication network. In another embodiment, one or more client nodes 160, server 110, and remote storage device 150 are located on different networks. In some embodiments, server 110 is located in a cloud network. Although not shown, in various embodiments, one or more client nodes 160 may be laptops, desktop computers, microprocessor-based or programmable consumer electronics, network devices, mobile phones, smartphones, pagers, radio frequency (RF) devices, infrared (IR) devices, personal digital assistants (PDAs), set-top boxes, cameras, integrated devices combining at least two of the above devices, etc. Figure 1A The dPCR processing module 120 is shown to also operate the PCR instrument 102 fully or partially. One or more client nodes 160 may include PDAs, GPS devices, displays, interactive televisions, Internet Protocol (IP) phones, pagers, cellular phones, satellite phones, etc.

[0035] In some embodiments, one or more client nodes 160 include a processor (not shown), memory (not shown), and a copy of the automatic thresholding agent 130. In some embodiments, one or more client nodes 160 may include processing software instructions and / or hardware logic required for automated PCR thresholding analysis according to embodiments described herein.

[0036] In some examples, server 110 provides remote cloud storage capabilities via remote storage device 150 connected by network 140 for PCR analysis and associated various types of PCR strategies. Additionally, server 110 provides remote storage functionality for PCR analysis data. In some embodiments, server 110 retrieves previous results and associated strategies related to the PCR analysis data from remote data storage 150 to local data storage 118. In other embodiments, databases of PCR automatic thresholding strategies, previous test results, etc., may be locally stored on one or more client nodes 160, local storage device 116, or server 110. Specifically, for remote storage purposes, local data storage unit 118 may be one or more centralized databases having a corresponding association mapping between each data fragment and its location within remote storage device 150. Local data storage device 116 may represent a single or multiple data structures (database, repository, file, etc.) located on one or more mass storage devices, such as magnetic or optical storage-based disks, tapes, or hard disks. This local data storage device 116 may be an internal component of server 110. Alternatively, the local data storage device 116 may be externally connected to the server 110, or remotely connected via a second network (not shown). Furthermore, the server 110 may communicate with the remote storage device via a public or private network. Although not shown, in various embodiments, the server 110 may be a laptop computer, desktop computer, microprocessor-based or programmable consumer electronics, network equipment, mobile phone, smartphone, radio frequency (RF) device, infrared (IR) device, personal digital assistant (PDA), set-top box, integrated device combining at least two of the above devices, etc.

[0037] Figure 1B This is a block diagram of an automatic thresholding agent 130 of server 110 according to some embodiments of FIG1. ​​Automatic thresholding agent 130 includes cluster detection agent 134, memory 131, processor 132, and storage device 133. Cluster detection agent 134 includes receiver 135, GMM engine 136, and K-means clustering engine 137. Cluster detection agent 134, cooperating with processor 132, is coupled to memory 131 and storage device 133 to retrieve and process automatic thresholding instructions. These automatic thresholding instructions may be stored in storage device 133. Receiver 135 is coupled to receive data from dPCR instrument 102 (…). Figure 1AThe receiver 135 is coupled to a GMM engine 136 and a K-means clustering engine 137 to transmit fluorescence data for detecting the presence of one or more clusters and generating thresholds based on the detected clusters.

[0038] In operation, a user of the dPCR system 100, which features improved automated thresholding performance and accuracy, opens the front panel to insert a microplate (not shown) into the dPCR instrument 102. Prior to placement, the user fills the microcompartment array of the microplate with a biological sample for dPCR analysis. A receiver 135 directs at least one memory 131 of the automated thresholding agent 130 (containing computer-readable instructions that, when executed by at least one processor 132) to receive fluorescence data from the dPCR instrument 102 corresponding to multiple optical signals detected on the dPCR instrument. The fluorescence data includes multiple individual points associated with multiple microcompartments, wherein these individual points form one or more clusters sharing relative characteristics. A GMM engine 136 is coupled to the receiver 135 and at least one memory 131, which contains relevant instructions enabling at least one processor 132 to use the GMM to detect whether the fluorescence data (e.g., one or more clusters) are associated with two different groups. Specifically, the GMM engine 136 detects clusters in fluorescence data by calculating the mean and standard deviation associated with the lower and higher GMM groups. The GMM engine 136 uses the following formula to calculate the thresholds for single and double groups, as well as the effective resolution:

[0039]

[0040]

[0041]

[0042]

[0043]

[0044] GMM thlo is the threshold of the lower group in GMM, ml is the mean of the lower group in GMM, and sl is the standard deviation of the lower group in GMM; GMM thhi Here, mh is the threshold of the higher-order group in the GMM, sh is the mean of the higher-order group in the GMM, and SD is the standard deviation of the higher-order group in the GMM. resolutionIt is GMM resolution, and abs represents the absolute value.

[0045] For cases where two clusters are detected using GMM, the GMM engine 136, cooperating with processor 132, specifies a bi-cluster GMM threshold as the final threshold. For example, if the resolution of the GMM algorithm is greater than threshold 1 = 0.7, then GMM engine 136 determines the existence of two clusters and uses the GMM bi-cluster threshold as the final threshold. In cases where two distinct clusters are not detected using GMM, the K-means clustering engine 137, cooperating with processor 132, is operable to use K-means clustering to detect whether the number of one or more clusters is related to two clusters in a bi-cluster group. Specifically, K-means clustering engine 137 detects clusters in the fluorescence data by calculating the mean and standard deviation associated with the lower and higher groups. K-means clustering engine 137 calculates the thresholds for one and two groups, as well as the effective resolution, using the following formula:

[0046]

[0047]

[0048]

[0049] Where Kmeans thlo is the threshold for the group with the lowest K-mean, ml is the mean of the group with the lowest K-mean, and sl is the standard deviation of the group with the lowest K-mean; Kmeans thhi is the threshold for the group with higher K-mean, mh is the mean of the group with higher K-mean, sh is the standard deviation of the group with higher K-mean, SD is the standard deviation of the whole (dataset), and abs represents the absolute value.

[0050] For cases where two clusters are detected using K-means clustering, the K-means clustering engine 137 specifies the bigroup K-means clustering threshold as the final threshold. In one example, when the K-means resolution is greater than 1.5, the K-means clustering engine 137 sets the threshold based on the K-means bigroup threshold. For cases where two clusters are not detected using K-means clustering, the K-means clustering engine 137 specifies the unigroup GMM threshold as the final threshold.

[0051] In some embodiments, server 110 retrieves previous dPCR analysis results and automatic threshold data from local storage units 116 and / or 118. In other embodiments, server 110 obtains previous dPCR analysis results and other parameter versions from remote storage 150. After performing the automatic thresholding function using dPCR processing module 120, server 110 sends the dPCR analysis result data to client node 160. The display 108 of client node 160 displays the PCR analysis result data in graphical, text, and various other formats, enabling users to review and further analyze it.

[0052] As described above regarding one example, the resolution (res1) derived by GMM must be greater than the threshold of 0.7 so that GMM engine 136 can determine the existence of two clusters. In some examples, the resolution (res2) derived by K-means clustering must be greater than the threshold of 1.5 so that K-means clustering engine 137 can determine the existence of two clusters. In another example, the overall standard deviation derived using GMM must be greater than the maximum negative threshold of 2900.00 in order to detect a cluster and calculate the associated GMM threshold. Such alternatives do not depart from the scope of the invention.

[0053] Figure 2AThis is a flowchart of an automated thresholding method for a dPCR operating system according to some embodiments of Figure 1. The automated thresholding method 200 includes receiving fluorescence data corresponding to multiple optical signals detected at a dPCR instrument, wherein the fluorescence data includes multiple individual points associated with multiple microchambers in operation 202. One or more clusters of the multiple individual points are data points sharing a relative characteristic. For example, the fluorescence data points of a cluster may indicate the presence of partial DNA or a pathogen. That is, the shared relative characteristic is associated with the target group described herein or its absence. In operation 210, the automated thresholding method 200 includes using a Gaussian mixture model (GMM) to detect whether one or more clusters are two clusters associated with two different groups. For example, a GMM engine 136 is coupled to a receiver 135 and at least one memory 131 having instructions to cause at least one processor 132 to use the GMM to detect whether one or more clusters are two clusters associated with two different groups. In decision operation 230, method 200 includes using the GMM to detect the presence of two different groups. For the case where two clusters are detected using GMM, the automatic thresholding method 200 further includes specifying a bi-cluster GMM threshold as the final threshold in operation 235. For the case where two distinct clusters are not detected using GMM, method 200 includes using K-means clustering in operation 240 to detect whether the number of one or more clusters is related to the two clusters in the bi-cluster cluster. For example, a K-means clustering engine 137 cooperating with processor 132 uses K-means clustering to detect whether the number of one or more clusters is related to the two clusters in the bi-cluster cluster. In decision operation 250, method 200 includes using K-means clustering to detect whether two distinct clusters exist. For the case where two clusters are detected using K-means clustering, the automatic thresholding method 200 includes specifying a bi-cluster K-means clustering threshold as the final threshold in operation 255. For the case where two clusters are not detected using K-means clustering, the automatic thresholding method 200 includes specifying a single-cluster GMM threshold as the final threshold in operation 260.

[0054] Figure 2B According to some embodiments Figure 2A Applications Figure 2AA flowchart of a GMM fitting method 210 using automatic thresholding for site statistics is provided. In operation 212, the GMM fitting method 210 includes calculating a single-group threshold (GMM_th1). In operation 220, the GMM fitting method 210 includes calculating a two-group threshold (GMM_th2). In operation 225, the GMM fitting method 210 includes calculating a first resolution (res1). In one example, the resolution (res1) is equal to the difference between the mean of the lower GMM group and the higher GMM group divided by twice the sum of the standard deviations of the lower GMM group and the higher GMM group ((ml-mh) / (2*(sl+sh))).

[0055] Figure 2C Applications according to some embodiments Figure 2A The flowchart illustrates the K-means clustering fit method 240 for site statistics using the automatic thresholding method 200. In operation 242, the K-means clustering fit method 240 includes calculating a two-group threshold (Kmeans_th2). In operation 246, the K-means clustering method 240 includes calculating a second resolution (res2). In one example, the resolution (res2) is equal to the difference between the means of the lower K-means group and the higher K-means group divided by twice the sum of the standard deviations of the lower K-means group and the higher K-means group ((ml-mh) / (2*(sl+sh))).

[0056] Figure 2D Based on the use of some embodiments Figure 2B The flowchart illustrates method 212 for calculating a single-group threshold using the GMM with automatic thresholding. In operation 214, method 212 includes calculating the mean and standard deviation (SD) of the population. In decision operation 215, method 212 includes detecting whether the standard deviation is greater than the maximum negative threshold. In another example, the population standard deviation derived using the GMM must be greater than the maximum negative threshold of 2900.00 in order to detect a group and calculate the associated GMM threshold. In operation 216, given that the standard deviation is less than or equal to the maximum negative value, the GMM threshold for a group is equal to the sum of the mean and four times the standard deviation (GMM_th1 = mean + 4 * SD). In operation 218, given that the standard deviation is greater than the maximum negative value, the GMM threshold for a group is equal to the difference between the mean and four times the standard deviation (GMM_th1 = mean + 4 * SD).

[0057] Figure 2E Based on the use of some embodiments Figure 2BThe flowchart of the method 220 for calculating the GMM threshold for the two groups in the automatic thresholding method 210 is provided. In operation 222, the GMM fitting method 210 includes calculating the mean and standard deviation (SD) of the lower and higher GMM groups. Specifically, the variables related to the GMM fitting include the mean (ml) of the lower GMM group, the mean (mh) of the higher GMM group, the standard deviation (sl) of the lower group, the standard deviation (sh) of the higher GMM group, the mean-mean of the entire dataset, and the standard deviation (SD) of the entire dataset. In operation 223, the GMM fitting method 210 includes calculating the GMM thresholds for the lower and higher GMM groups. Specifically, the GMM threshold for the lower group is equal to the sum of the mean and standard deviation of the lower group (gmm_th_lo = ml + sl); however, the GMM threshold for the higher GMM group is equal to the difference between the mean and standard deviation of the higher GMM group (gmm_th_hi = mh - sh). In operation 223, the GMM fitting method 210 includes calculating the GMM threshold for the bigroup response. For example, the bigroup GMM threshold is equal to the average of the GMM thresholds for the lower and higher GMM groups (e.g., GMM_th2 = (gmm_th_lo + gmm_th_hi) / 2).

[0058] Figure 2F Based on the use of some embodiments Figure 2CThe flowchart illustrates the method 242 for calculating the threshold of two groups using K-means clustering in the automatic thresholding method 240. In operation 243, the K-means clustering fitting method 240 includes calculating the mean and standard deviation (SD) of the lower K-mean group and the higher K-mean group (ml, mh, sl, sh). In operation 244, the K-means clustering fitting method 240 includes calculating the K-means clustering threshold for the lower K-mean group and the higher K-mean group. Specifically, the K-means threshold for the lower K-mean group is equal to the sum of the mean and the standard deviation of the lower K-mean group (means_th_lo = ml + sl); while the K-means threshold for the higher K-mean group is equal to the difference between the mean and the standard deviation of the higher K-mean group (kmeans_th_hi = mh - sh). In operation 245, the K-means clustering fitting method 240 includes calculating the K-means clustering threshold for the two-group responses. For example, the K-means clustering threshold for the two groups is equal to the average of the K-means clustering thresholds for the lower and higher K-means groups (e.g., kmeans_th2 = (kmeans_th_lo + kmeans_th_hi) / 2).

[0059] Now for reference Figure 3Figure 3 shows an example of a computing device that can implement the examples described herein. In this example, computer system 300 may provide one or more components of dPCR / analysis configured to implement one or more logical modules and artificial neural networks, as well as related components for computer-implemented dPCR analysis and associated interactive graphical user interfaces. It should be understood that the methods described herein can be performed using digital processing systems, such as conventional general-purpose computer systems. Special-purpose computers, which are specifically designed or programmed to perform a single function, may also be used. The computing device of Figure 3 can be used to perform embodiments of functions for performing dPCR analysis according to some embodiments. The computing device includes a central processing unit (CUP) 302 coupled to memory 304 and mass storage device 308 via bus 306. Mass storage device 308 represents a permanent data storage device, such as a floppy disk drive or fixed disk drive, which may be local or remote in some embodiments. In some embodiments, mass storage device 308 may implement backup storage. Memory 304 may include read-only memory, random access memory, etc. In some embodiments, applications residing on a computing device may be stored on or accessed through a computer-readable medium such as memory 304 or mass storage device 308. Applications may also be in the form of modulated electronic signals accessed via a network modem or other network interface of the computing device. It should be understood that in some embodiments, the central processing unit 302 may be embodied as a general-purpose processor, a dedicated processor, or specially programmed logic means.

[0060] Display 312 communicates with central processing unit 302, memory 304, and mass storage device 308 via bus 306. Display 312 is configured to display any visualization tools or reports related to the system described herein. Input / output device 310 is coupled to bus 306 to transmit information from command selection to central processing unit 302. It should be understood that data to and from external devices can be transmitted via input / output device 310. Central processing unit 302 can be defined to perform the functions described herein to implement the functions described with reference to Figures 1A-2F. In some embodiments, code embodying this function can be stored in memory 304 or mass storage device 308 for execution by a processor such as central processing unit 302. The operating system on the computing device may be iOS. TM MS-WINDOWS TM OS / 2 TM UNIX TM LINUX TMOr other known operating systems. It should be understood that the embodiments described herein can also be integrated with virtualized computing systems.

[0061] Many details have been set forth in the foregoing description. However, those skilled in the art will understand that the present invention can be practiced without relying on these specific details. In some instances, to avoid obscuring the spirit of the invention, some of the accompanying drawings show known structures and devices in block diagram form rather than in detail.

[0062] It should be understood that the above description is intended to be illustrative and not restrictive. Many other embodiments will be apparent to those skilled in the art upon reading and understanding the above description. While the invention has been described with reference to specific exemplary embodiments, it should be recognized that the invention is not limited to the described embodiments, but can be practiced with modifications and variations within the spirit and scope of the appended claims. Therefore, the specification and drawings should be considered illustrative and not restrictive. Consequently, the scope of protection of the invention should be determined by reference to the full scope of the appended claims and their legal equivalents.

[0063] This document discloses detailed exemplary embodiments. However, the specific functional details disclosed herein are for the purpose of describing representative embodiments only. Furthermore, embodiments may be implemented in various alternative forms and should not be construed as being limited to the schemes set forth herein.

[0064] It should be understood that although the terms first, second, etc., may be used herein to describe various steps or calculations, these steps or calculations should not be limited by these terms. These terms are used only to distinguish different steps or calculations. For example, without departing from the scope of this disclosure, a first calculation may be referred to as a second calculation, and similarly, a second step may be referred to as a first step. As used herein, the terms “and / or” and the “ / ” symbol include any one and all combinations of one or more of the listed related items. Unless the context clearly indicates otherwise, the singular forms “a / an” and “the” are intended to include the plural forms as used herein. It should also be understood that when the terms “comprising,” “including,” “containing,” and / or “having” are used herein, they indicate the presence of the stated feature, integral, step, operation, element, and / or component, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or combinations thereof. Therefore, the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting.

[0065] It should also be noted that in some alternative embodiments, the mentioned functions / operations may not occur in the order shown in the figures. For example, two figures shown consecutively may actually be performed substantially simultaneously, or sometimes in reverse order depending on the functions / behaviors involved. Considering the above embodiments, it should be understood that the embodiments may employ various computer operations involving data stored in a computer system. These operations are precisely those processes that require physical manipulation of physical quantities. Typically, but not necessarily, these quantities take the form of electrical or magnetic signals that can be stored, transferred, combined, compared, and otherwise manipulated. Furthermore, these performed operations are generally referred to using terms such as generating, identifying, determining, or comparing. Any operations described herein that form part of the embodiments are useful machine operations. The embodiments described herein also relate to means or apparatus for performing these operations. The means may be specifically constructed for the desired purpose, or the means may be a general-purpose computer selectively activated or configured by a computer program stored in a computer. Specifically, various general-purpose machines programmed with computer programs according to the teachings described herein may be used, or it may be more convenient to construct more specialized means to perform the desired operations.

[0066] Modules, applications, layers, agents, or other method-operable entities may be implemented as hardware, firmware, or a processor executing software, or a combination thereof. It should be understood that, in the software-based embodiments disclosed herein, the software may be embodied in a physical machine such as a controller. For example, a controller may include a first module and a second module. The controller may be configured to perform various operations, such as the operation of a method, application, layer, or agent.

[0067] Some embodiments may also be implemented as computer-readable code on a non-transitory computer-readable medium. A computer-readable medium is any data storage device capable of storing data that can subsequently be read by a computer system. Examples of computer-readable media include hard disk drives, network-attached storage devices, read-only memory, random access memory, CD-ROMs, CD-Rs, CD-RWs, magnetic tape, flash memory devices, and other optical and non-optical data storage devices. The computer-readable medium may also be distributed across networked computer systems, allowing the computer-readable code to be stored and executed in a distributed manner. The embodiments described herein are applicable to a variety of computer system configurations, including handheld devices, tablets, microprocessor systems, microprocessor-based programmable consumer electronics, minicomputers, mainframes, etc. This embodiment is also applicable to distributed computing environments where tasks are performed by remote processing devices connected via wired or wireless networks.

[0068] Although the method operations are described in a specific order, it should be understood that additional operations may be inserted between the operations, the timing of the operations may be adjusted to make them slightly different, or the operations may be assigned to a system that allows the processing operations to occur at different time intervals related to the processing.

[0069] In various embodiments, one or more portions of the methods and mechanisms described herein may form part of a cloud computing environment. In such embodiments, resources may be provided as a service over the Internet according to one or more various models. Such models may include Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and Software as a Service (SaaS). In IaaS, computing infrastructure is delivered as a service. In this case, computing devices are typically owned and operated by the service provider. In the PaaS model, software tools and underlying devices used by developers to develop software solutions may be provided as a service and hosted by the service provider. SaaS typically refers to a service provider licensing software services on demand. The service provider may self-host the software or deploy it to the customer side for a specific period of time. Combined applications of the above models are all feasible and under consideration.

[0070] Various units, circuits, or other components may be described or declared as "configured to" or "operable to" perform one or more tasks. In this context, the phrase "configured to" implies a structure that includes a unit / line / component that performs one or more tasks during operation (e.g., a circuit). Thus, even if the specified device / circuit / component is not currently operational (e.g., not turned on), it can be considered configured to perform a task. Units / circuit / components described using "configured to" include hardware; for example, circuits, memory storing program instructions executable to perform that operation, etc. The statement that a unit / circuit / component is "configured to" perform one or more tasks is expressly intended to exclude the application of 35 USC 112, paragraph 6, for that unit / circuit / component. Furthermore, "configured to" may include general-purpose structures (e.g., general-purpose circuits) that are operated by software and / or firmware (e.g., a field-programmable gate array (FPGA) or a general-purpose processor executing software) in a manner capable of performing tasks. "Configure as" may also include adjusting the manufacturing process (e.g., a semiconductor manufacturing plant) to manufacture a device (e.g., an integrated circuit) suitable for performing one or more tasks.

[0071] Within the context of this specification, at least the following embodiments are described. Embodiment 1 is a method for processing digital polymerase chain reaction (dPCR) data from biological samples. The method includes receiving fluorescence data corresponding to multiple light signals detected by a dPCR instrument, wherein the fluorescence data corresponds to the fluorescence intensity of multiple sample partitions; determining whether the fluorescence data is correlated with two different groups using a first mathematical model; and, for a result where the first mathematical model determines that the fluorescence data is not correlated with the two different groups, determining whether the fluorescence data is correlated with the two different groups using a second mathematical model. Embodiment 2 is the method of Embodiment 1, further comprising: for cases where the first mathematical model determines that the fluorescence data is correlated with two different groups, specifying a two-group threshold determined by the first mathematical model as a final threshold.

[0072] Example 3 is the method of any one of Examples 1 and 2, further comprising: for cases where the fluorescence data is determined to be correlated with two different groups using a second mathematical model, designating the bigroup threshold determined by the second mathematical model as the final threshold. Example 4 is the method of any one of Examples 1 to 3, further comprising: for cases where the fluorescence data is determined to be uncorrelated with two different groups using a second mathematical model, designating the single-group threshold determined by the first mathematical model as the final threshold.

[0073] Example 5 is the method of any one of Examples 1 to 4, wherein the first mathematical model includes a Gaussian Mixture Model (GMM). Example 6 is the method of Example 5, wherein the second mathematical model includes a K-means clustering model. Example 7 is the method of Example 6, wherein determining whether fluorescence data is associated with two different groups using GMM includes defining a GMM fit for a higher GMM group and a lower GMM group; calculating a first threshold associated with a first cluster based on the GMM fit for the higher GMM group and the lower GMM group; calculating a second threshold associated with a second cluster based on the GMM fit for the higher GMM group and the lower GMM group; determining the GMM resolution for the first cluster and the second cluster; and identifying the fluorescence data as associated with two different groups if the GMM resolution is greater than a first predetermined variable.

[0074] Example 8 describes the method described in Example 7, wherein defining the GMM fitting for the first and second clusters includes calculating the mean of the higher GMM group, the mean of the lower GMM group, and the mean of the entire group including all multiple partitions; calculating the standard deviation of the entire group, the standard deviation of the higher GMM group, and the standard deviation of the lower GMM group; setting the GMM threshold associated with the lower GMM group as the sum of the mean and standard deviation of the lower GMM group; and setting the GMM threshold associated with the higher GMM group as the difference between the mean and standard deviation of the higher GMM group.

[0075] Example 9 is the method described in Example 8, wherein, for cases where GMM-based fluorescence data are associated with two different groups, setting the final threshold for the fluorescence data includes calculating the sum of the GMM threshold for the higher GMM group and the GMM threshold for the lower GMM group; dividing the sum by 2 to obtain a first average value as the final threshold.

[0076] Example 10 describes the method described in Example 9, wherein, for cases where fluorescence data based on K-means clustering is unrelated to two different groups, GMM is used to obtain the final threshold through the following steps: detecting whether the difference between the mean of the entire group and the standard deviation of the entire group is greater than a second predetermined variable; if the detected difference is greater than the second predetermined variable, setting the final threshold of a single cluster to the difference between the mean of the entire group and four times the standard deviation of the entire group; and if the detected difference is less than the predetermined variable, setting the final threshold of the fluorescence data to the sum of the mean of the entire group and four times the standard deviation of the entire group.

[0077] Example 11 is the method described in any one of Examples 6 to 10, wherein using K-means clustering to detect whether fluorescence data is associated with two different groups includes defining a K-means fit for a group with a higher K-mean and a group with a lower K-mean; calculating a first K-means threshold associated with a first cluster based on the K-means fit for the group with the higher K-mean and the group with the lower K-mean; calculating a second K-means threshold associated with a second cluster based on the K-means fit for the group with the higher K-mean and the group with the lower K-mean; determining the K-means resolution for the first cluster and the second cluster; identifying the fluorescence data as associated with two different groups if the K-means resolution is greater than or equal to a third predetermined variable; and identifying the fluorescence data as unassociated with two different groups if the K-means resolution is less than the third predetermined variable.

[0078] Example 12 describes the method described in Example 11, wherein defining the K-mean fitting of fluorescence data associated with two different groups includes calculating the mean of the higher K-mean group, wherein the higher K-mean group is a cluster representing sample partitions where the target analyte is present; calculating the mean of the lower K-mean group, wherein the lower K-mean group is a cluster representing sample partitions where the target analyte is absent; calculating the standard deviation of the higher K-mean group and the standard deviation of the lower K-mean group; setting the K-mean threshold for the higher K-mean group as the difference between the mean of the higher K-mean group and the standard deviation of the higher K-mean group; and setting the K-mean threshold for the lower K-mean group as the sum of the mean of the lower K-mean group and the standard deviation of the lower group.

[0079] Example 13 describes the method described in Example 12, wherein, for cases where fluorescence data is determined to be associated with two different groups based on K-means clustering, the sum of the K-mean threshold of the group with higher K-mean and the K-mean threshold of the group with lower K-mean is calculated; the sum is divided by 2 to obtain a second average value; and the K-mean threshold associated with the two-cluster group is set as the second average value to obtain the final threshold of the fluorescence data.

[0080] Example 14 is the method of any one of Examples 11 to 13, wherein the K-mean resolution is calculated as follows: the absolute value of the difference between the mean of the lower K-mean group and the mean of the higher K-mean group, divided by twice the sum of the standard deviations of the lower K-mean group and the higher K-mean group. Example 14 is the method of any one of Examples 2 to 14, wherein data points with fluorescence data above the final threshold correspond to sample partitions containing the target analyte, and data points with fluorescence data below the final threshold correspond to sample partitions without the target analyte. Example 16 is the method of Example 15, wherein the target analyte comprises a nucleic acid molecule or nucleic acid fragment. Example 17 is the automated thresholding method of any one of Examples 2 to 16, wherein the final threshold is configured to facilitate the quantification of the target analyte.

[0081] Example 18 is a non-transitory computer-readable medium comprising instructions executable by one or more computer processors to implement the method according to any one of claims 1 to 17.

[0082] Example 19 is an automated thresholding system for data obtained using a dPCR instrument, the automated thresholding system including a memory; and a processor configured to receive fluorescence data corresponding to multiple light signals detected by the dPCR instrument, wherein the fluorescence data corresponds to the fluorescence intensity of multiple sample partitions; using a first mathematical model including a Gaussian mixture model (GMM) to determine whether the fluorescence data is associated with two different groups; and for cases where the first mathematical model determines that the fluorescence data is not associated with the two different groups, using a second mathematical model including K-means clustering to determine whether the fluorescence data is associated with the two different groups.

[0083] Example 20 is the automatic thresholding system described in Example 19, wherein determining whether fluorescence data is associated with two different groups using a first mathematical model includes defining a GMM fitting of multiple data points into a higher GMM group and a lower GMM group; calculating a first threshold associated with the higher GMM group based on the higher GMM group and the lower GMM group; calculating a second threshold associated with the lower GMM group based on the higher GMM group and the lower GMM group; determining the GMM resolution of the first cluster and the second cluster; identifying multiple data points as associated with the two different groups if the GMM resolution is greater than a first predetermined variable; and identifying multiple data points as unassociated with the two different groups if the GMM resolution is not greater than the first predetermined variable.

[0084] Example 21 is the automatic thresholding system described in Example 20, wherein the GMM resolution is calculated as follows: the absolute value of the difference between the average value of the lower group of GMM and the average value of the higher group of GMM, divided by twice the sum of the standard deviations of the lower group of GMM and the standard deviations of the higher group of GMM. Example 22 is an automatic thresholding system as described in any one of Examples 19 to 21, wherein determining whether fluorescence data is associated with two different groups using K-mean clustering includes defining K-mean fits for a higher K-mean group and a lower K-mean group; calculating a first K-mean threshold associated with a first cluster based on the K-mean fits for the higher K-mean group and the lower K-mean group; calculating a second K-mean threshold associated with a second cluster based on the K-mean fits for the higher K-mean group and the lower K-mean group; determining the K-mean resolution of the first cluster and the second cluster; identifying fluorescence data as associated with two different groups when the K-mean resolution is greater than or equal to a third predetermined variable; and identifying fluorescence data as unassociated with two different groups when the K-mean resolution is less than the third predetermined variable.

[0085] Example 23 describes the method of Example 22, wherein the K-mean resolution is calculated as follows: the absolute value of the difference between the mean of the lower K-mean group and the mean of the higher K-mean group is divided by twice the sum of the standard deviations of the lower K-mean group and the higher K-mean group. Example 24 describes the automatic thresholding system of any one of claims 19 to 23, wherein, for cases where GMM is used to determine the correlation between fluorescence data and two different groups, the processor is further configured to specify the bigroup threshold determined by the GMM as the final threshold. Example 25 describes the automatic thresholding system of claim 24, wherein the bigroup threshold determined by the GMM is obtained by dividing the sum of the GMM thresholds of the higher and lower GMM groups by 2.

[0086] Example 26 is an automatic thresholding system according to any one of Examples 19 to 25, wherein, for cases where fluorescence data is determined to be associated with two different groups using K-means clustering, the processor is further configured to specify the bi-group threshold determined by K-means clustering as the final threshold. Example 27 is an automatic thresholding system according to Example 23, wherein the bi-group threshold determined by K-means clustering is obtained by dividing the sum of the K-mean threshold of the group with the higher K-means and the K-mean threshold of the group with the lower K-means by 2. Example 28 is an automatic thresholding system according to any one of Examples 19 to 27, wherein, for cases where fluorescence data is determined to be unrelated to two different groups using K-means clustering, the processor is further configured to specify the single-group threshold determined by GMM as the final threshold.

[0087] Example 29 is the automatic thresholding system described in Example 28, wherein the threshold of a single group determined by the GMM is obtained by detecting whether the difference between the mean of the entire group and the standard deviation of the entire group is greater than a second predetermined variable; if the difference is greater than the second predetermined variable, the final threshold of the fluorescence data is set to the difference between the mean of the entire group and four times the standard deviation of the entire group; and if the difference is less than the predetermined variable, the final threshold of the fluorescence data is set to the sum of the mean of the entire group and four times the standard deviation of the entire group.

[0088] Example 30 is the automated thresholding system described in any one of Examples 24 to 29, wherein data points with fluorescence data above a final threshold correspond to sample partitions containing the target analyte, and data points with fluorescence data below the final threshold correspond to sample partitions without the target analyte. Example 31 is the automated thresholding system described in Example 30, wherein the target analyte comprises a nucleic acid molecule or nucleic acid fragment. Example 32 is the automated thresholding system described in any one of Examples 24 to 31, wherein the final threshold is configured to facilitate the quantification of the target analyte. Example 33 is a digital polymerase chain reaction (dPCR) system comprising a server, having at least one processor coupled to at least one memory, and a dPCR processing module including an automatic thresholding system. The automatic thresholding system includes a cluster detection unit having a receiver coupled to the at least one memory, the at least one memory storing computer-readable instructions that, when executed by the at least one processor, cause the at least one processor to receive fluorescence data from a dPCR instrument corresponding to multiple light signals detected by the dPCR instrument; wherein the fluorescence data includes multiple intensity values ​​corresponding to multiple sample partitions; a Gaussian mixture model (GMM) engine, coupled to the receiver and the at least one memory storing computer-readable instructions, that, when the at least one processor executes the instructions, causes the at least one processor to use a Gaussian mixture model (GMM) to detect whether the fluorescence data are two clusters associated with two different groups; for the case where two clusters are detected using GMM, a bi-group GMM threshold is specified as the final threshold; and a K-means clustering engine, coupled to the GMM engine and the at least one memory storing computer-readable instructions, that, when executed by the at least one processor, for the case where two clusters are detected using GMM... If no two different groups are detected, the at least one processor uses K-means clustering to detect whether the fluorescence data are two clusters related to the two different groups; if two clusters are detected using K-means clustering, the bi-group K-means threshold is specified as the final threshold; and if two clusters are not detected using K-means clustering, the single-group GMM threshold is specified as the final threshold.

[0089] For illustrative purposes, the foregoing has been described with reference to specific embodiments. However, the exemplary discussion above is not exhaustive and is not intended to limit the invention to the precise forms disclosed. In view of the foregoing teachings, many modifications and variations of the invention are possible. The specific embodiments selected and described herein are intended to clearly illustrate the technical principles of the embodiments and their practical applications, enabling those skilled in the art to best utilize these embodiments and make appropriate adjustments according to specific application scenarios. Therefore, the embodiments described herein should be considered illustrative rather than restrictive, and the invention should not be limited to the details described herein, but can be modified within the scope of the appended claims and their equivalents.

Claims

1. A method for processing digital polymerase chain reaction (dPCR) data from biological samples, the method comprising: Receive fluorescence data corresponding to multiple light signals detected by the dPCR instrument, wherein the fluorescence data corresponds to the fluorescence intensity of multiple sample partitions; The first mathematical model was used to determine whether the fluorescence data were correlated with two different groups; as well as For the results obtained by using the first mathematical model to determine that the fluorescence data are not correlated with the two different groups, the second mathematical model is used to determine whether the fluorescence data are correlated with the two different groups.

2. The method of claim 1, further comprising: For cases where the fluorescence data is determined to be associated with two different groups using the first mathematical model, the bigroup threshold determined by the first mathematical model is specified as the final threshold.

3. The method according to any one of claims 1 and 2, further comprising: For cases where the fluorescence data is determined to be associated with two different groups using a second mathematical model, the bigroup threshold determined by the second mathematical model is specified as the final threshold.

4. The method according to any one of claims 1 to 3, further comprising: For cases where the fluorescence data is determined to be uncorrelated with two different groups using the second mathematical model, the single-group threshold determined by the first mathematical model is specified as the final threshold.

5. The method according to any one of claims 1 to 4, wherein the first mathematical model comprises a Gaussian mixture model (GMM).

6. The method according to claim 5, wherein the second mathematical model comprises a K-means clustering model.

7. The method of claim 6, wherein using GMM to determine whether the fluorescence data is associated with two different groups comprises: Define the GMM fit for the higher and lower groups of the GMM; Based on the GMM fitting of the higher and lower groups of the GMM, calculate the first threshold associated with the first cluster; Based on the GMM fitting of the higher and lower GMM groups, a second threshold associated with the second cluster is calculated; Determine the GMM resolution for the first and second clusters; as well as For cases where the GMM resolution is greater than the first predetermined variable, fluorescence data are identified as being associated with two different groups.

8. The method of claim 7, wherein defining GMM fitting for the first cluster and the second cluster comprises: Calculate the average of the higher-order GMM groups, the average of the lower-order GMM groups, and the average of the entire group (including all multiple partitions); Calculate the standard deviation of the entire group, the standard deviation of the higher group in the GMM, and the standard deviation of the lower group in the GMM; Set the GMM threshold associated with the lower GMM group as the sum of the lower GMM group mean and the lower GMM group standard deviation; as well as The GMM threshold associated with the higher GMM group is set as the difference between the mean of the higher GMM group and the standard deviation of the higher GMM group.

9. The method of claim 8, wherein setting a final threshold for the fluorescence data in cases where the fluorescence data determined by GMM is associated with two different groups includes: Calculate the sum of the GMM thresholds for the higher group and the lower group; and Divide the sum by 2 to obtain the first average value, which is used as the final threshold.

10. The method of claim 9, wherein, for cases where fluorescence data is determined to be unrelated to two different groups based on K-means clustering, the final threshold is obtained using GMM through the following steps: Detect whether the difference between the mean and the standard deviation of the entire group is greater than the second predetermined variable; For cases where the detected difference is greater than the second predetermined variable, the final threshold for a single cluster is set as the difference between the mean of the entire cluster and four times the standard deviation of the entire cluster. as well as For cases where the detected difference is less than a predetermined variable, the final threshold for fluorescence data is set as the sum of the mean of the entire group and four times the standard deviation of the entire group.

11. The method according to any one of claims 6 to 10, wherein using K-means clustering to detect whether the fluorescence data is associated with two different groups comprises: Define the K-mean fit for the group with higher K-mean and the group with lower K-mean; Based on the K-mean fitting of the high K-mean group and the low K-mean group, calculate the first K-mean threshold associated with the first cluster; Based on the K-mean fitting of the high K-mean group and the low K-mean group, calculate the second K-mean threshold associated with the second cluster; Determine the K-mean resolution of the first and second clusters; For cases where the K-mean resolution is greater than or equal to the third predetermined variable, the fluorescence data are determined to be correlated with two distinct groups. as well as For cases where the K-mean resolution is less than that of the third predetermined variable, it was determined that the fluorescence data were independent of the two different groups.

12. The method of claim 11, wherein defining the K-means fit of fluorescence data associated with two different groups comprises: Calculate the mean of the K-means higher cluster, where the K-means higher cluster represents the sample partitions containing the target analyte; Calculate the mean of the low K-mean group, where the low K-mean group is a cluster representing a partition where no target analyte is present; Calculate the standard deviation of the group with higher K-mean and the standard deviation of the group with lower K-mean; Set the K-mean threshold for the group with higher K-mean to the difference between the mean of the group with higher K-mean and the standard deviation of the group with higher K-mean. as well as Set the K-mean threshold for the group with the lower K-mean to the sum of the mean of the group with the lower K-mean and the standard deviation of the group with the lower K-mean.

13. The method of claim 12, wherein, for the case where fluorescence data is determined to be associated with two different groups based on K-means clustering, the final threshold of the fluorescence data is obtained through the following steps: Calculate the sum of the K-mean thresholds for the group with higher K-means and the K-mean thresholds for the group with lower K-means; Divide the sum by 2 to get the second average; and Set the K-mean threshold associated with the bi-cluster group to the second mean.

14. The method according to any one of claims 11 to 13, wherein the K-mean resolution is calculated by dividing the absolute value of the difference between the mean of the lower K-mean group and the mean of the higher K-mean group by twice the sum of the standard deviations of the lower K-mean group and the higher K-mean group.

15. The method according to any one of claims 2 to 14, wherein data points with fluorescence data above a final threshold correspond to sample partitions containing the target analyte, and data points with fluorescence data below the final threshold correspond to sample partitions without the target analyte.

16. The method of claim 15, wherein the target analyte comprises a nucleic acid molecule or a nucleic acid fragment.

17. The method according to any one of claims 2 to 16, wherein the final threshold is configured to facilitate the quantification of the target analyte.

18. A non-transitory computer-readable medium comprising instructions executable by one or more computer processors to implement the method according to any one of claims 1 to 17.

19. An automated thresholding system for data obtained using dPCR instruments, including: Memory; as well as Processor, the processor being configured to: Receive fluorescence data corresponding to multiple light signals detected by the dPCR instrument, wherein the fluorescence data corresponds to the fluorescence intensity of multiple sample partitions; The first mathematical model, including the Gaussian mixture model (GMM), was used to determine whether the fluorescence data were associated with two different groups; as well as For cases where the first mathematical model determines that the fluorescence data is not correlated with the two different groups, a second mathematical model, including K-means clustering, is used to determine whether the fluorescence data is correlated with the two different groups.

20. The automatic thresholding system of claim 19, wherein determining whether the fluorescence data is associated with two different groups using the first mathematical model includes: Define the GMM fitting for the higher and lower groups of GMMs for multiple data points; Calculate the first threshold associated with the higher GMM group based on the higher GMM group and the lower GMM group; Based on the higher and lower GMM groups, calculate the second threshold associated with the lower GMM group; Determine the GMM resolution for the first and second clusters; as well as For cases where the GMM resolution is greater than the first predetermined variable, multiple data points are identified as being associated with two different groups; as well as For cases where the GMM resolution is no greater than the first predetermined variable, multiple data points are determined to be uncorrelated with two different groups.

21. The automatic thresholding system according to claim 20, wherein the GMM resolution is calculated as follows: the absolute value of the difference between the mean of the lower group of GMM and the mean of the higher group of GMM, divided by twice the sum of the standard deviations of the lower group of GMM and the standard deviations of the higher group of GMM.

22. The automatic thresholding system according to any one of claims 19 to 21, wherein using K-means clustering to determine whether the fluorescence data is associated with two different groups includes: Define the K-mean fit for the group with higher K-mean and the group with lower K-mean; Based on the K-mean fitting of the high K-mean group and the low K-mean group, calculate the first K-mean threshold associated with the first cluster; Based on the K-mean fitting of the high K-mean group and the low K-mean group, calculate the second K-mean threshold associated with the second cluster; Determine the K-mean resolution of the first and second clusters; For cases where the K-mean resolution is greater than or equal to the third predetermined variable, the fluorescence data are determined to be correlated with two distinct groups. as well as For cases where the K-mean resolution is less than that of the third predetermined variable, it was determined that the fluorescence data were independent of the two different groups.

23. The automatic thresholding system according to claim 22, wherein the K-mean resolution is calculated as follows: the absolute value of the difference between the mean of the lower K-mean group and the mean of the higher K-mean group is divided by twice the sum of the standard deviations of the lower K-mean group and the standard deviations of the higher K-mean group.

24. The automatic thresholding system according to any one of claims 19 to 23, wherein, for cases where fluorescence data is determined to be associated with two different groups using a GMM, the processor is further configured to designate the dual-group threshold determined by the GMM as the final threshold.

25. The automatic thresholding system of claim 24, wherein the bigroup threshold determined by GMM is obtained by dividing the sum of the GMM threshold of the higher group and the GMM threshold of the lower group by 2.

26. The automatic thresholding system according to any one of claims 19 to 25, wherein, for cases where K-means clustering is used to determine that fluorescence data is associated with two different groups, the processor is further configured to designate the bi-group threshold determined by K-means clustering as the final threshold.

27. The automatic thresholding system of claim 23, wherein the bi-group threshold determined by K-mean clustering is obtained by dividing the sum of the K-mean thresholds of the higher K-mean group and the lower K-mean group by 2.

28. The automatic thresholding system according to any one of claims 19 to 27, wherein, for cases where K-means clustering is used to determine that fluorescence data is not correlated with two different groups, the processor is further configured to designate a single-group threshold determined by the GMM as the final threshold.

29. The automatic thresholding system of claim 28, wherein the single-group threshold determined by the GMM is obtained in the following manner: Detect whether the difference between the mean and the standard deviation of the entire group is greater than the second predetermined variable; For cases where the difference exceeds the second predetermined variable, the final threshold for fluorescence data is set as the difference between the mean of the entire population and four times the standard deviation of the entire population; and For cases where the detected difference is less than a predetermined variable, the final threshold for fluorescence data is set as the sum of the mean of the entire group and four times the standard deviation of the entire group.

30. The automatic thresholding system according to any one of claims 24 to 29, wherein data points with fluorescence data above a final threshold correspond to sample partitions containing the target analyte, and data points with fluorescence data below the final threshold correspond to sample partitions without the target analyte.

31. The automated thresholding system of claim 30, wherein the target analyte comprises a nucleic acid molecule or a nucleic acid fragment.

32. The automated thresholding system according to any one of claims 24 to 31, wherein the final threshold is configured to facilitate the quantification of the target analyte.

33. A digital polymerase chain reaction (qPCR) system, comprising: A server includes at least one processor coupled to at least one memory and a dPCR processing module including an automatic thresholding system, wherein the automatic thresholding system includes: A cluster detection unit includes a receiver coupled to at least one memory storing computer-readable instructions thereon, which, when executed by at least one processor, causes at least one processor to: Receive fluorescence data corresponding to multiple optical signals detected by the dPCR instrument; The fluorescence data includes multiple intensity values ​​corresponding to multiple sample zones; The Gaussian Mixture Model (GMM) engine is coupled to a receiver and at least one memory thereon storing computer-readable instructions, which, when executed by at least one processor, causes at least one processor to: A Gaussian mixture model (GMM) was used to determine whether the fluorescence data belonged to two clusters associated with two different groups. For cases where two clusters are detected using GMM, the dual-group GMM threshold is specified as the final threshold; The K-means clustering engine is coupled to the GMM engine and at least one memory thereon storing computer-readable instructions, which, when executed by the at least one processor, cause the at least one processor to: In cases where GMM fails to detect two distinct groups, K-means clustering is used to determine whether the fluorescence data represent two clusters related to the two distinct groups. For cases where two clusters are detected using K-means clustering, the bigroup K-means threshold is specified as the final threshold; and For cases where two clusters are not detected using K-means clustering, the single-group GMM threshold is specified as the final threshold.

Images