Base identification method and device, equipment and storage medium

By obtaining and using the correct and misclassification information and contextual sequence characteristics of the initial base recognition results of the sequencing wheel, the base recognition results are calibrated, and the problem of high base recognition error rate in the DNA sequencing platform is solved, improving the accuracy of the sequencing data.

CN118942549APending Publication Date: 2024-11-12GENEMIND BIOSCIENCES CO LTD
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202410540991.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-04-30
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

The existing DNA sequencing platform has a high error rate in the base recognition process, which affects the accuracy of the sequencing data and the downstream bioinformatics analysis results.

Method used

By obtaining the mapping relationship between the correct and error classification information of the initial base recognition result of the sequencing wheel and the first sequencing information, based on the characteristics of the target sequencing wheel and its context sequence, the base recognition results of the pending sequencing wheel are calibrated to reduce sequencing errors based on the characteristics of the target sequencing wheel and its context sequence, the base recognition results of the pending sequencing wheel are calibrated to reduce sequencing errors Rate.

Benefits of technology

It significantly improves the accuracy of base recognition, reduces the impact of high-frequency errors, and improves the quality of sequencing data.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118942549A_ABST
    Figure CN118942549A_ABST
Patent Text Reader

Abstract

The invention discloses a base recognition method, device and equipment and a storage medium, and the method comprises the steps: obtaining a first mapping relation between correct and wrong classification information of an initial base recognition result of a sequencing round and first sequencing information, and the first sequencing information comprises first initial base recognition information based on a specified sequencing round; and on the basis of the first mapping relationship and the first sequencing information of the to-be-processed sequencing wheel, determining the correct and wrong classification information of the initial base identification result of the to-be-processed sequencing wheel. According to the base recognition method provided by the invention, the influence of the context sequence on base recognition is reduced, and the accuracy of base recognition is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of biological information processing technology, and in particular to a base recognition method, device, equipment and storage medium. Background Art

[0002] DNA sequencing is one of the most basic technologies in modern life sciences. The high-quality sequencing data provided by DNA sequencing technology depends on accurate base recognition during the sequencing process. However, all sequencing platforms have a certain base recognition error rate, which may affect the downstream bioinformatics analysis results and thus affect the accuracy of the corresponding research results. Therefore, for any sequencing platform, reducing the error rate is the key to improving sequencing quality. Therefore, how to improve the accuracy of base recognition has become an important research direction. Summary of the invention

[0003] The present invention provides a base recognition method, device, equipment and storage medium to improve the accuracy of base recognition.

[0004] According to one aspect of the present invention, there is provided a base recognition method, comprising:

[0005] Obtaining a first mapping relationship between correct and incorrect classification information of an initial base recognition result of a sequencing round and first sequencing information, wherein the first sequencing information includes first initial base recognition information based on a target sequencing round;

[0006] Based on the first mapping relationship and the first sequencing information of the sequencing round to be processed, correct or incorrect classification information of the initial base recognition result of the sequencing round to be processed is determined.

[0007] According to another aspect of the present invention, there is provided a base recognition device, comprising:

[0008] A first mapping relationship acquisition module is configured to acquire a first mapping relationship between the correct and incorrect classification information of the initial base recognition result of the sequencing round and the first sequencing information, wherein the first sequencing information includes the first initial base recognition information based on the target sequencing round, and the first initial base recognition information includes the initial base recognition information of a consecutive sequencing rounds, and the a consecutive sequencing rounds at least include the target sequencing round;

[0009] The right-or-wrong classification information determination module is used to determine the right-or-wrong classification information of the initial base recognition result of the sequencing round to be processed based on the first mapping relationship and the first sequencing information of the sequencing round to be processed.

[0010] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0011] at least one processor; and

[0012] a memory communicatively connected to at least one processor; wherein,

[0013] The memory stores a computer program that can be executed by at least one processor. The computer program is executed by at least one processor so that the at least one processor can perform the method of any embodiment of the present invention.

[0014] According to another aspect of the present invention, a computer-readable storage medium is provided. The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement a method of any embodiment of the present invention when executed.

[0015] The technical solution of the embodiment of the present invention obtains the first mapping relationship between the correct and incorrect classification information of the initial base recognition result of the sequencing round and the first sequencing information, wherein the first sequencing information includes the first initial base recognition information based on the target sequencing round; based on the first mapping relationship and the first sequencing information of the sequencing round to be processed, the correct and incorrect classification information of the initial base recognition result of the sequencing round to be processed is determined. Thus, by using the base recognition information of a consecutive sequencing round to calibrate the base recognition information of the sequencing round to be processed, especially by using the characteristics of the target sequencing round and its context sequence, the sequencing error rate can be significantly reduced, the impact of high-frequency errors can be effectively reduced, and the accuracy of base recognition can be improved.

[0016] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0018] Figure 1 It is a schematic diagram of a flow chart of a base recognition method provided by an embodiment of the present invention;

[0019] Figure 2 is a schematic structural diagram of a base recognition device provided by an embodiment of the present invention;

[0020] Figure 3 is a schematic structural diagram of another base recognition device provided by an embodiment of the present invention;

[0021] Figure 4is a schematic structural diagram of another base recognition device provided by an embodiment of the present invention;

[0022] Figure 5 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0024] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein.

[0025] The term "sequencing" can also be called "nucleic acid sequencing" or "gene sequencing", that is, the three can be expressed interchangeably, referring to the determination of base types and arrangement orders in nucleic acid sequences; including synthesis sequencing (sequencing by synthesis, SBS) and / or ligation sequencing (sequencing by ligation, SBL), including DNA sequencing and / or RNA sequencing, including long fragment sequencing and / or short fragment sequencing. The so-called long fragments and short fragments are relative, such as nucleic acid molecules longer than 1Kb, 2Kb, 5Kb or 10Kb can be called long fragments, and those shorter than 1Kb or 800bp can be called short fragments.

[0026] Sequencing generally includes multiple rounds of sequencing to achieve the determination of the order of multiple nucleotides / bases on the template: "One round of sequencing" (cycle) is also called a "sequencing round", which can be defined as a base extension of four nucleotides / bases, that is, it can be defined as the determination of the base type at any specified position on the nucleic acid template. For sequencing platforms that achieve sequencing based on polymerization reactions or ligation reactions, one round of sequencing includes the process of achieving one-time binding of four nucleotides (including nucleotide analogs) to the corresponding nucleic acid template, and collecting the corresponding signals generated by the four nucleotides (including nucleotide analogs) after binding. Generally, one round of sequencing may include one or more base extensions (repeat), for example, four nucleotides are added to the reaction system in sequence, and base extensions and corresponding reaction signals are collected respectively. In this case, one round of sequencing includes four base extensions; for another example, four nucleotides are added to the reaction system in any combination, such as two-by-two combinations or one-three combinations, and two combinations are respectively subjected to base extension and corresponding reaction signal collection. In this case, one round of sequencing includes two base extensions; for another example, four nucleotides are added to the reaction system at the same time for base extension and reaction signal collection. In this case, one round of sequencing includes one base extension.

[0027] Sequencing can be performed through a sequencing platform. According to the embodiments of the present application, the available sequencing platforms include, but are not limited to, Illumina's Hiseq, Miseq, Nextseq, and Novasek sequencing platforms, Thermo Fisher / Life Technologies' Ion Torrent platform, BGISEQ, MGISEQ / DNBSEQ platforms, and single-molecule sequencing platforms.

[0028] In the description herein, A represents adenine, and may also represent adenine nucleotide or an analog thereof; C represents cytosine, and may also represent cytosine nucleotide or an analog thereof; G represents guanine, and may also represent guanine nucleotide or an analog thereof; T represents thymine, and may also represent thymine nucleotide or an analog thereof; U represents uracil, and may also represent uracil nucleotide or an analog thereof. It should be understood that in the embodiments of the present invention, the expressions of A, C, G and T / U are consistent, and when one of them represents a base, the other three are also represented as bases, such as: when A represents adenine, correspondingly, C represents cytosine, G represents guanine, T represents thymine / U represents uracil; when one of them represents a nucleotide or an analog thereof, the other three are also represented as nucleotides or an analog thereof, such as: when A can represent adenine nucleotide or an analog thereof, correspondingly, C represents cytosine nucleotide or an analog thereof, G represents guanine nucleotide or an analog thereof, T represents thymine nucleotide or an analog thereof / U represents uracil or an analog thereof. The “ / ” in T / U means “or”, that is, “T / U” means “T or U”.

[0029] In the description of this article, unless otherwise clearly defined, based on image information, the so-called "intensity" and pixel (pixel value) can be replaced, and the so-called intensity or pixel size can be a real or objective absolute value, or a relative value, including various deformations based on the real pixel value, such as an enlarged pixel value, a reduced pixel value, a ratio or relationship based on pixel values, etc. Generally, when comparing the intensity / pixel size of multiple images or bright spots or positions, the intensity / pixel size of these images or bright spots or positions is the intensity / pixel size after the same processing, such as objective pixel values ​​or pixel values ​​after the same deformation processing; when comparing and analyzing the information of specific positions based on one or more images, when determining these specific positions, it is preferred that these images are aligned and located in the same coordinate system. In one embodiment, the "intensity" referred to in the embodiment of the present invention may be "fluorescence intensity".

[0030] The so-called "spots" (spots or peaks) on the image, also called "bright spots" or "light spots", refer to the positions on the image where the signal is relatively strong, for example, the signal at the position is stronger than the surrounding ones, and it appears as a relatively bright spot or point on the image. A bright spot or a so-called position occupies one or more pixels. The signal of the bright spot / position may come from the target molecule or from non-target substances. The detection of "bright spots" includes the detection of the optical signal of the target molecule such as the extended base or base cluster.

[0031] The term "crosstalk" (laser-crosstalk or spectra-crosstalk), also known as "spectral crosstalk" or "spectral crossing", refers to the phenomenon that the signal corresponding to one base diffuses into the signal of another base; for sequencing platforms that use differently labeled fluorescent molecules to identify different bases, if the emission spectra of two or more selected fluorescent molecules overlap, it may be detected that the signal of one fluorescent molecule diffuses into another fluorescent channel in one round of sequencing.

[0032] The term "reaction asynchrony" is also called "phase dephasing", "phase imbalance", "dephasing", and "phase difference". It refers to the phenomenon of asynchronous reactions between nucleic acid molecules in a group, such as a nucleic acid molecule cluster, in a chemical reaction, including phasing or sequence lag and prephasing or sequence lead; in a sequencing platform that uses differently labeled fluorescent molecules to identify different bases, it manifests as the phenomenon that the signal of the fluorescent molecule corresponding to the base at a specific position is not zero in more than one round of sequencing. Generally, sequencing is performed using nucleotides labeled with fluorescent molecules and blocking groups. The blocking group on the nucleotide can prevent other nucleotides from binding to the next position of the template. The blocking group is, for example, an azide at the 3' position of the sugar group of the nucleotide. The shedding of the blocking group or its failure to be removed before the extension of the next base will cause phase dephasing.

[0033] The term "channel" refers to four channels formed in different ways during the sequencing process that can screen and distinguish the four bases A, C, G, T (or U). For example, the so-called channel can refer to four optical channels of fluorescent signals that can screen and distinguish the four fluorescent bases A, C, G, T (or U) formed by using different excitation lights, different fluorescent filters, etc. during the sequencing process. During actual sequencing, photos will be taken in four different fluorescent channels to form an image. Ideally, each fluorescent channel only has the signal of the fluorescent base category corresponding to the channel, but in reality, due to the influence of fluorescent crosstalk, in addition to the fluorescent signal of the corresponding fluorescent base, each channel will also have fluorescent signals of other bases.

[0034] The term "base recognition error rate" means: the number of bases that are incorrectly recognized (which can be confirmed by comparison with a standard reference genome) is the ratio of N2 to the total number of bases recognized N1. The "base recognition error rate" is represented by P, then P = N2 / N1.

[0035] In addition, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus that includes a series of steps or elements is not necessarily limited to those steps or elements explicitly listed, but may include other steps or elements not explicitly listed or inherent to such process, method, product, or apparatus.

[0036] Figure 1It is a flow chart of a base recognition method provided by an embodiment of the present invention. This embodiment can be applied to the situation when the base recognition result is determined according to the sequencing information of the sequencing round. The method can be executed by a base recognition device. The base recognition device can be implemented in the form of hardware and / or software. The base recognition device can be configured in an electronic device.

[0037] like Figure 1 As shown, the method includes:

[0038] S110: Obtain a first mapping relationship between correct and incorrect classification information of an initial base recognition result of a sequencing round and first sequencing information.

[0039] In the embodiment of the present application, the first mapping relationship between the correct and incorrect classification information of the initial base recognition result of the sequencing round and the first sequencing information refers to the relationship that the first sequencing information based on the sequencing round can correspond to the correct and incorrect classification information of the initial base recognition result of the sequencing round.

[0040] It should be understood that in the embodiments of the present application, the "sequencing round" in the "correction classification information of the initial base recognition result of the sequencing round" refers to the sequencing round pointed to by the feature (first sequencing information) and the target (correction classification information of the initial base recognition result). Unless otherwise specified, a sequencing round generally refers to a sequencing round, and in some embodiments, the sequencing round can also be understood as a target sequencing round, or a designated sequencing round.

[0041] In the embodiment of the present application, the initial base recognition result refers to the initial result of the sequencing platform identifying the base type of the target sequencing round, i.e., the target sequencing round, according to the existing base recognition software when the sequencing round is subjected to the base extension reaction. The output form of the initial result can be a base type, such as A base, T or U base, G base and C base; or it can be the probability score of each base type being detected or recognized, such as the probability score of A base 94%, the probability score of T or U base 2%, the probability score of G base 2% and the probability score of C base 2%. It should be understood that in conventional base recognition software, as a general understanding, if not otherwise specified, the base type of the sequencing round output by the base recognition software can be determined by the probability of each base type. Specifically, the base with the highest probability among the four bases is considered to be the base type recognized by the base recognition software.

[0042] In the embodiments of the present application, the correct and incorrect classification information of the initial base recognition result refers to the information for judging the correct classification and incorrect classification of the initial base recognition result. In some embodiments, the correct and incorrect classification information of the initial base recognition result includes the correct classification and incorrect classification of the identified base type. In other embodiments, the correct and incorrect classification information of the initial base recognition result includes the correct classification of the identified base type accuracy and the probability score of the incorrect classification. It should be understood that, as a general understanding, if not otherwise specified, the correct classification and incorrect classification can be determined by the probability score of the base obtained by the base recognition software.

[0043] In some embodiments, the correct and incorrect classification information can be a correct and incorrect category, including a correct category and an incorrect category. In some embodiments, the correct and incorrect category can be a binary classification, i.e., a correct category and an incorrect category. In one implementation scenario, if the initial base recognition result of the sequencing round base extension is consistent with the true base alignment, the initial base recognition of the sequencing round is considered to be correct and is classified as the correct category in the correct and incorrect classification information. Exemplarily, the initial base recognition result of the sequencing round base extension is A, and the reagent base type is also A, then the initial base recognition of the sequencing round is considered to be correct, and at this time, it is classified as the correct category in the correct and incorrect classification information. In another implementation scenario, if the initial base recognition result of the sequencing round base extension is inconsistent with the true base alignment, then the initial base recognition of the sequencing round is considered to be wrong, and it is classified as the wrong category in the correct and incorrect classification information. Exemplarily, the initial base recognition result of the sequencing round base extension is A, and the true base is T, then the initial base recognition of the sequencing round is considered to be wrong, and at this time, it is classified as the wrong category in the correct and incorrect classification information.

[0044] In some embodiments, the error categories may be further classified according to different types of errors, so that the right and wrong categories are multiple categories with the number of categories being greater than 2. In one embodiment, the right and wrong categories are five categories, and the five categories include the following five categories:

[0045] The correct category of the initial base calling result;

[0046] A base error category;

[0047] T or U base error category;

[0048] G base error category;

[0049] C base error category.

[0050] The classification basis of the A base error category is: the actual base corresponding to the sequencing round is A, but the initial base recognition result of the sequencing round base extension is not A. At this time, it is considered that the correct error type of the initial base recognition of the sequencing round is: A base error category. In some embodiments, the A base error category can be further classified according to the initial base recognition result of the sequencing round base extension, illustratively, the category of A base error recognition as T, the category of A base error recognition as G, and the category of A base error recognition as C.

[0051] Similarly, the classification basis of the T or U base error category is: the actual base corresponding to the sequencing round is T, but the initial base recognition result of the sequencing round base extension is not T. At this time, it is considered that the correct error type of the initial base recognition of the sequencing round is: T or U base error category. In some embodiments, the T or U base error category can be further classified according to the initial base recognition result of the sequencing round base extension, illustratively, the T or U base error is recognized as the category of A, the T or U base error is recognized as the category of G, and the T or U base error is recognized as the category of C.

[0052] The classification basis of the G base error category is: the actual base corresponding to the sequencing round is G, but the initial base recognition result of the sequencing round base extension is not G. At this time, it is considered that the correct error type of the initial base recognition of the sequencing round is: G base error category. In some embodiments, the G base error category can be further classified according to the initial base recognition result of the sequencing round base extension, illustratively, the category of G base error recognition as T, the category of G base error recognition as A, and the category of G base error recognition as C.

[0053] The classification basis of the C base error category is: the actual base corresponding to the sequencing round is C, but the initial base recognition result of the sequencing round base extension is not C. At this time, it is considered that the correct error type of the initial base recognition of the sequencing round is: C base error category. In some embodiments, the C base error category can be further classified according to the initial base recognition result of the sequencing round base extension, illustratively, the category of C base error recognition as A, the category of C base error recognition as G, and the category of C base error recognition as T.

[0054] In the embodiment of the present application, the first sequencing information includes the first initial base recognition information of the sequencing round that requires base recognition, that is, the target sequencing round.

[0055] Considering that when the target sequencing round is base identified, the identifiable features may be affected by factors such as the context base type and its base extension reaction. Therefore, when constructing the mapping relationship between the correct and incorrect classification information of the initial base recognition result of the target sequencing round and the first sequencing information, the corresponding relationship between the base recognition information of the context sequence of the target sequencing round base and the standard base of the target sequencing round is considered, so that the recognition of the base can take into account the influencing factors of the "context sequence", and based on the obtained error characteristics related to the base to be identified and the context sequence, the errors related to the context sequence of the base to be identified in the sequencing are corrected to achieve the effect of reducing the sequencing error rate, while reducing the high-frequency errors in the above-mentioned specific context environment, and finally improving the accuracy of base recognition. It should be understood that the "context sequence" referred to in the embodiment of the present application refers to the base sequence obtained by completing the base extension reaction before the target sequencing round, i.e., the "previous sequence", and the base sequence obtained by performing the base extension reaction after the base extension reaction is completed in the target sequencing round, i.e., the "below sequence".

[0056] In view of this, the first initial base identification information includes initial base identification information of a consecutive sequencing rounds, and the a consecutive sequencing rounds at least include a target sequencing round.

[0057] In the embodiment of the present application, the consecutive a sequencing rounds include a target sequencing round, and at least one of the following two situations:

[0058] n1 sequencing rounds that perform base extension reactions prior to the target sequencing round,

[0059] m1 sequencing rounds in which base extension reaction is performed later than the target sequencing round.

[0060] That is, a consecutive sequencing rounds include three implementation scenarios:

[0061] In a first implementation scenario, the a consecutive sequencing rounds include a target sequencing round and n1 sequencing rounds in which base extension reactions occur before the target sequencing round, ie, the first n1 sequencing rounds. In this case, a=n1+1.

[0062] In the second implementation scenario, the consecutive a sequencing rounds include a target sequencing round and m1 sequencing rounds in which base extension reactions occur after the target sequencing round, ie, the last m1 sequencing rounds. In this case, a=m1+1.

[0063] In the third implementation scenario, the a consecutive sequencing rounds include the target sequencing round, n1 sequencing rounds in which base extension reactions occur before the target sequencing round, i.e., the first n1 sequencing rounds, and m1 sequencing rounds in which base extension reactions occur after the target sequencing round, i.e., the last m1 sequencing rounds. In this case, a=n1+m1+1, and a is a natural number greater than or equal to 2. This implementation scenario can take into account the following influences on base recognition in the sequencing rounds, and can improve the accuracy of base recognition to a greater extent.

[0064] Among them, the values ​​of m1 and n1 satisfy: n1 is an integer greater than or equal to 0, and m1 is an integer greater than or equal to 0. When m1 and n1 are both 0, a consecutive sequencing rounds are the target sequencing rounds, that is, the sequencing rounds that require base recognition. At this time, the first initial base recognition information sequence only contains the photoelectric signal generated from the target sequencing round. Therefore, in an embodiment of the present application, m1 and n1 are not 0 at the same time, that is: at least one of m1 and n1 is not 0. Exemplarily, at least one of m1 and n1 is not 0, including: m1 is 0, n1 is not 0; m1 is not 0, n1 is 0; m1 is not 0, and n1 is not 0.

[0065] In some embodiments, m1 and n1 are each a natural number from 1 to 50. For example, m1 can be 1, 2, 3, 4, 5, 8, 10, 12, 15, 18, 20, 22, 25, 28, 30, 32, 35, 38, 40, 42, 45, 48, 50, etc., and n1 can be 1, 2, 3, 4, 5, 8, 10, 12, 15, 18, 20, 22, 25, 28, 30, 32, 35, 38, 40, 42, 45, 48, 50, etc. It should be understood that the above values ​​are only examples. In fact, the values ​​of m1 and n1 are not limited thereto.

[0066] It should be understood that when a consecutive sequencing rounds simultaneously include n1 sequencing rounds in which base extension reactions occur before the target sequencing round, i.e., the first n1 sequencing rounds, and m1 sequencing rounds in which base extension reactions occur after the target sequencing round, i.e., the last m1 sequencing rounds, there is no strict requirement for the selection of m1 and n1, and they can be the same or different.

[0067] In the embodiment of the present application, the first initial base identification information can be selected in different ways, such as: the first initial base identification information can be the initial base identification information sequence of a consecutive sequencing rounds, or the initial base identification information of each of a consecutive sequencing rounds, or the initial base identification information sequence formed by some consecutive sequencing rounds among a consecutive sequencing rounds, and the initial base identification information of each of the remaining sequencing rounds.

[0068] In one implementation scenario, the first initial base identification information includes the initial base identification information sequence of a consecutive sequencing rounds, that is, the context sequence of the target sequencing round (including the target sequencing round). At this time, the first initial base identification information is a sequence formed by a consecutive base identification information. Exemplarily, the initial base identification information sequence can be the initial base identification information sequence of n1+1 consecutive sequencing rounds including the target sequencing round, or the initial base identification information sequence of m1+1 consecutive sequencing rounds including the target sequencing round, or the initial base identification information sequence of n1+m1+1 consecutive sequencing rounds including the target sequencing round. At this time, the initial base identification information sequence as a whole is associated with the correct and incorrect classification information of the initial base identification result of the sequencing round.

[0069] In another implementation scenario, the first initial base identification information includes at least one of the initial base identification information sequences of n1 sequencing rounds and the initial base identification information sequences of m1 sequencing rounds, and the initial base identification information of the target sequencing round. In some embodiments, the first initial base identification information may be the initial base identification information sequences of n1 consecutive sequencing rounds, and the initial base identification information of the target sequencing round. At this time, the initial base identification information sequences of n1 consecutive sequencing rounds, as a whole, together with the initial base identification information of the target sequencing round, are associated with the correct and incorrect classification information of the initial base identification results of the sequencing round. In some embodiments, the first initial base identification information may be the initial base identification information sequences of m1 consecutive sequencing rounds, and the initial base identification information of the target sequencing round. At this time, the initial base identification information sequences of m1 consecutive sequencing rounds, as a whole, together with the initial base identification information of the target sequencing round, are associated with the correct and incorrect classification information of the initial base identification results of the sequencing round. In some embodiments, the first initial base identification information may be a collection of the initial base identification information sequences of n1 consecutive sequencing rounds, the initial base identification information sequences of m1 consecutive sequencing rounds, and the initial base identification information of the target sequencing round. In this case, the initial base identification information sequences of n1 consecutive sequencing rounds as a whole, the initial base identification information sequences of m1 consecutive sequencing rounds as a whole, together with the initial base identification information of the target sequencing round, are associated with the correct or incorrect classification information of the initial base identification results of the sequencing rounds.

[0070] In another implementation scenario, the first initial base identification information includes the initial base identification information of each sequencing round in a consecutive sequencing rounds, that is, the set of initial base identification information of each of the consecutive a sequencing rounds. In some embodiments, the first initial base identification information is the initial base identification information of each sequencing round in n1 consecutive sequencing rounds, and the set formed by the initial base identification information of the target sequencing round. At this time, the initial base identification information of each of the n1+1 consecutive sequencing rounds is collected together to associate the correct and incorrect classification information of the initial base identification results of the sequencing round. In some embodiments, the first initial base identification information is the initial base identification information of each sequencing round in m1 consecutive sequencing rounds, and the set formed by the initial base identification information of the target sequencing round. At this time, the initial base identification information of each of the m1+1 consecutive sequencing rounds is collected together to associate the correct and incorrect classification information of the initial base identification results of the sequencing round. In some embodiments, the first initial base identification information is the initial base identification information of each sequencing round in n1 consecutive sequencing rounds, the initial base identification information of each sequencing round in m1 consecutive sequencing rounds, and the set formed by the initial base identification information of the target sequencing round. At this time, the initial base recognition information of each of the n1+m1+1 consecutive sequencing rounds is collected together to associate the correct and incorrect classification information of the initial base recognition results of the sequencing rounds.

[0071] In the embodiment of the present application, the first initial base recognition information refers to the initial information obtained when base recognition is performed on a number of consecutive sequencing rounds including the target sequencing round. The initial information can be the real or objective original data output by the base recognition, or it can be the relative data obtained after processing the original data output by the base recognition, but not as the final base recognition result.

[0072] In some embodiments, the first initial base identification information is the photoelectric signal intensity generated by the base extension reaction of a continuous sequencing round including the target sequencing round. In some embodiments, the photoelectric signal intensity includes the optical signal intensity of multiple optical signal channels, wherein the multiple optical signals correspond to the multiple base types participating in the base extension reaction. In this embodiment, there is a corresponding relationship between the optical signal and the base type and the optical signal channel. In one embodiment, each base type corresponds to an optical signal channel, and the signal generated by the type of base in the optical signal channel is collected to obtain a corresponding optical signal, such as an optical signal set of a specific wavelength. Exemplarily, based on the base type, the optical signal channel for collecting optical signals is four channels, including an A base channel, a C base channel, a G base channel, and a T / U base channel. In another embodiment, two or more base types obtain their respective optical signals through the same optical signal channel, and each signal in the optical signal channel is collected to obtain an optical signal corresponding to the number of base types, such as an optical signal set of two or more specific wavelengths, and each optical signal set contains all optical signals obtained at a specific wavelength. Exemplarily, based on the base type, the optical signal channel for collecting the optical signal is a dual channel, including an A / T / U base channel and a G / C base channel.

[0073] In some embodiments, the first initial base recognition information includes the optical signal intensity of each optical signal channel. Taking four channels as an example, the optical signal intensity from the A base channel, the C base channel, the G base channel and the T / U base channel is used as a component of the first initial base recognition information.

[0074] In some embodiments, the first initial base recognition information is the optical signal intensity with the largest intensity among the multiple optical signal channels. Taking four channels as an example, if the optical signal intensity in the A base channel is 0.9, and the optical signal intensities in the C base channel, the G base channel, and the T / U base channel are 0.2, 0.2, and 0.4, respectively, the optical signal intensity in the A base channel is used as a component of the first initial base recognition information.

[0075] In an embodiment in which a base fluorescence signal is used as a base recognition signal, the photoelectric signal intensity is the fluorescence signal intensity. At this time, the first initial base recognition information includes the fluorescence signal intensity of the fluorescence channel of a consecutive sequencing rounds including the target sequencing round. Wherein, based on the base type, the fluorescence channel may include an A base fluorescence channel, a C base fluorescence channel, a G base fluorescence channel, and a T / U base fluorescence channel.

[0076] In an embodiment of the present application, the optical signal intensity includes any one of the initial optical signal intensity and the corrected optical signal intensity. Among them, the initial optical signal intensity refers to the optical signal intensity determined directly from the image, and the optical signal intensity can be obtained by extracting the intensity of the optical signal of the bright spot or the effective bright spot in the image. The corrected optical signal intensity refers to the intensity obtained after the initial optical signal intensity is corrected by a preset correction method. The preset correction method includes background correction, crosstalk correction, and reaction asynchrony correction, etc. In addition, the normalization process can also be understood as a kind of correction method, that is, the correction intensity includes at least one of the background correction intensity, the crosstalk correction intensity, the reaction asynchrony correction intensity, and the intensity after normalization, but is not limited to this.

[0077] In some embodiments using base fluorescence signals as base recognition signals, the optical signal intensity may be: the fluorescence intensity directly extracted from the optical image collected by the optical imaging device, i.e., the initial fluorescence intensity; or the image of the base collected by the optical imaging device is processed, such as the relative fluorescence intensity obtained after background correction. Of course, the optical signal intensity is not limited to the above-mentioned cases, and may also be other data obtained after processing the above-mentioned cases, such as the fluorescence intensity.

[0078] In one embodiment, the first initial base identification information includes a sequence of calibrated intensities obtained from base extension reactions of a consecutive sequencing rounds including the target sequencing round. The situation of a consecutive sequencing rounds is as described above, and will not be repeated here to save space.

[0079] Exemplarily, the method for confirming the correction intensity includes: taking four channels as an example, collecting optical images obtained from the A base channel, C base channel, G base channel and T / U base channel in each sequencing round extension reaction, extracting the original optical signal intensity at each bright spot in the optical image; for each bright spot, the original optical signal intensity of the A base channel, C base channel, G base channel and T / U base channel is used as the initial intensity of the A base, C base, G base and T / U base respectively; performing at least one of background correction, crosstalk correction, reaction asynchrony correction and normalization processing on the initial intensity, and the correction intensity of the A base, C base, G base and T / U base can be determined. In this embodiment, the intensity after correction is used as the source of the first initial base recognition information, which can remove interference caused by factors such as background, crosstalk and reaction asynchrony, and improve the purity of the sequencing signal; when it is used as a feature value to input into the corresponding machine learning and / or deep learning model for machine learning and / or deep learning, it is more conducive to improving the accuracy of model training.

[0080] In the embodiment of the present application, the first mapping relationship between the correct and incorrect classification information of the initial base recognition result of the sequencing round and the first sequencing information can be a mapping mathematical model, a mapping neural network model, etc. The first mapping relationship can be obtained by taking the sequencing information of the target sequencing round in the standard sample as a feature value and the correct and incorrect classification information of the initial base recognition result corresponding to the target sequencing round of the standard sample as a target value, and inputting them into a pre-set machine learning and / or deep learning model for machine learning and / or deep learning, that is, the mapping relationship between the correct and incorrect classification information of the initial base recognition result of the sequencing round and the first sequencing information is obtained by training the pre-constructed base recognition result correct and incorrect classification model.

[0081] In some embodiments, the training of the base call result correct or incorrect classification model includes:

[0082] (11) obtaining a standard sample and a sequence of the standard sample;

[0083] (12) performing a base extension reaction on the standard sample to determine the second sequencing information of the target sequencing round and the initial base recognition result of the base extension of the target sequencing round, wherein the second sequencing information at least includes the first sequencing information;

[0084] (13) determining the correct or incorrect category of the initial base recognition of the target sequencing round based on the initial base recognition result of the target sequencing round base extension and the standard base;

[0085] (14) constructing a training sample using the second sequencing information as a feature value and the correct or incorrect category of the initial base recognition sequence as a target value;

[0086] (15) Based on the training samples, a pre-constructed base recognition result correct / error classification model is trained to obtain a trained base recognition result correct / error classification model.

[0087] In the above step (11), the standard sample refers to a nucleic acid sample of a known base sequence. In the embodiment of the present application, there is no strict requirement for the selection of the standard sample. It can be a human reference gene sequence or a reference gene sequence of other species. It only needs to satisfy the condition that the sequence is a known sequence. In some embodiments, a sample with a larger genome is selected as the standard sample; samples containing a variety of different species can also be used as standard samples together. Thus, by deriving the training data from a larger genome sample or a genome sample of a non-single species, the training data can cover more features, especially the features of the context sequence required by the embodiment of the present application, thereby making the base recognition model more generalizable. It should be understood that "features of the context sequence" refers to the sequencing information used for context base recognition, such as the first sequencing information, the second sequencing information, and the like described above.

[0088] In the embodiments of the present application, the sequence of the standard sample can be directly obtained from the information provided by the standard sample, or the sequence of the standard sample can be determined by comparing it with the reference genome after sequencing the standard sample. Exemplarily, the standard sample is sequenced on a selected sequencing platform to obtain the corresponding base sequence; the obtained base sequence is compared with the reference genome of the standard sample to obtain a known sequence.

[0089] In the above step (12), a base extension reaction is performed on the standard sample to obtain the second sequencing information and the initial base recognition result of the target sequencing round. In the embodiment of the present application, the initial base recognition result of the target sequencing round refers to the result of performing a base extension reaction on the standard sample, and the sequencing platform recognizes the base type of the target sequencing round according to the existing base recognition software, and the result is the base type, such as A base, T or U base, G base and C base.

[0090] In an embodiment of the present application, the second sequencing information is information associated with base recognition when the target sequencing round of the standard sample performs a base extension reaction. In theory, the wider the information range covered by the second sequencing information, the more features covered by the training data, the better the generalization of the base recognition model, and the higher the calibration accuracy of the base recognition model. It should be understood that the second sequencing information at least includes the first sequencing information in the embodiment of the present application, that is, the first sequencing information is derived from the second sequencing information. In one embodiment, the first sequencing information is the same as the second sequencing information. In an embodiment of the present application, the second sequencing information at least includes the first initial base recognition information of a continuous sequencing round obtained by performing a base extension reaction on the standard sample, and a continuous sequencing round includes at least the target sequencing round. Among them, the initial base recognition information of a continuous sequencing round can be referred to the above description, and it will not be described here in order to save space.

[0091] In some embodiments, the first initial base identification information is the photoelectric signal intensity generated during the base extension reaction. In some embodiments, the photoelectric signal intensity includes the optical signal intensity of multiple optical signal channels, wherein the multiple optical signals correspond to the multiple base types participating in the base extension reaction. In some embodiments, the optical signal intensity is the optical signal intensity with the largest intensity in multiple optical signal channels. In some embodiments, the optical signal intensity is the fluorescence signal intensity. In some embodiments, the fluorescence signal intensity includes any one of the initial fluorescence signal intensity and the corrected fluorescence signal intensity. In some embodiments, the correction includes at least one of background correction, crosstalk correction, reaction asynchrony correction, and normalization. In order to save space, it will not be repeated here.

[0092] In one implementation scenario, the first initial base identification information includes the initial base identification information sequence of a consecutive sequencing rounds, that is, the context sequence of the target sequencing round (including the target sequencing round). At this time, the first initial base identification information is a sequence formed by a consecutive base identification information. Exemplarily, the initial base identification information sequence can be the initial base identification information sequence of n1+1 consecutive sequencing rounds including the target sequencing round, or the initial base identification information sequence of m1+1 consecutive sequencing rounds including the target sequencing round, or the initial base identification information sequence of n1+m1+1 consecutive sequencing rounds including the target sequencing round. At this time, the initial base identification information sequence as a whole is associated with the correct and incorrect classification information of the initial base identification result of the sequencing round.

[0093] In another implementation scenario, the first initial base identification information includes at least one of the initial base identification information sequences of n1 sequencing rounds and the initial base identification information sequences of m1 sequencing rounds, and the initial base identification information of the target sequencing round. In some embodiments, the first initial base identification information may be the initial base identification information sequences of n1 consecutive sequencing rounds, and the initial base identification information of the target sequencing round. At this time, the initial base identification information sequences of n1 consecutive sequencing rounds, as a whole, together with the initial base identification information of the target sequencing round, are associated with the correct and incorrect classification information of the initial base identification results of the sequencing round. In some embodiments, the first initial base identification information may be the initial base identification information sequences of m1 consecutive sequencing rounds, and the initial base identification information of the target sequencing round. At this time, the initial base identification information sequences of m1 consecutive sequencing rounds, as a whole, together with the initial base identification information of the target sequencing round, are associated with the correct and incorrect classification information of the initial base identification results of the sequencing round. In some embodiments, the first initial base identification information may be a collection of the initial base identification information sequences of n1 consecutive sequencing rounds, the initial base identification information sequences of m1 consecutive sequencing rounds, and the initial base identification information of the target sequencing round. In this case, the initial base identification information sequences of n1 consecutive sequencing rounds as a whole, the initial base identification information sequences of m1 consecutive sequencing rounds as a whole, together with the initial base identification information of the target sequencing round, are associated with the correct or incorrect classification information of the initial base identification results of the sequencing rounds.

[0094] In another implementation scenario, the first initial base identification information includes the initial base identification information of each sequencing round in a consecutive sequencing rounds, that is, the set of initial base identification information of each of the consecutive a sequencing rounds. In some embodiments, the first initial base identification information is the initial base identification information of each sequencing round in n1 consecutive sequencing rounds, and the set formed by the initial base identification information of the target sequencing round. At this time, the initial base identification information of each of the n1+1 consecutive sequencing rounds is collected together to associate the correct and incorrect classification information of the initial base identification results of the sequencing round. In some embodiments, the first initial base identification information is the initial base identification information of each sequencing round in m1 consecutive sequencing rounds, and the set formed by the initial base identification information of the target sequencing round. At this time, the initial base identification information of each of the m1+1 consecutive sequencing rounds is collected together to associate the correct and incorrect classification information of the initial base identification results of the sequencing round. In some embodiments, the first initial base identification information is the initial base identification information of each sequencing round in n1 consecutive sequencing rounds, the initial base identification information of each sequencing round in m1 consecutive sequencing rounds, and the set formed by the initial base identification information of the target sequencing round. At this time, the initial base recognition information of each of the n1+m1+1 consecutive sequencing rounds is collected together to associate the correct and incorrect classification information of the initial base recognition results of the sequencing rounds.

[0095] In the above step (13), the initial base recognition result based on the base extension of the target sequencing round is compared with the corresponding standard base in the standard sample sequence of the target sequencing round to determine the correct or incorrect category of the initial base recognition of the target sequencing round.

[0096] In an embodiment of the present application, the correct and incorrect categories include at least two categories, correct and incorrect. In some embodiments, the correct and incorrect categories can be two categories, namely, correct categories and incorrect categories. In one implementation scenario, if the initial base recognition result of the target sequencing round base extension is consistent with the standard base alignment corresponding to the target sequencing round in the standard sample sequence, the initial base recognition of the target sequencing round is considered to be correct. Exemplarily, the initial base recognition result of the target sequencing round base extension is A, and the standard base corresponding to the target sequencing round in the standard sample sequence is also A, then the initial base recognition of the target sequencing round is considered to be correct. In another implementation scenario, if the initial base recognition result of the target sequencing round base extension is inconsistent with the standard base alignment corresponding to the target sequencing round in the standard sample sequence, then the initial base recognition of the target sequencing round is considered to be wrong. Exemplarily, the initial base recognition result of the target sequencing round base extension is A, and the standard base corresponding to the target sequencing round in the standard sample sequence is also T, then the initial base recognition of the target sequencing round is considered to be wrong.

[0097] In some embodiments, the error categories may be further classified according to different types of errors, so that the right and wrong categories are multiple categories with the number of categories being greater than 2. In one embodiment, the right and wrong categories are five categories, and the five categories include the following five categories:

[0098] The correct category of the initial base calling result;

[0099] A base error category;

[0100] T or U base error category;

[0101] G base error category;

[0102] C base error category.

[0103] The classification basis of the A base error category is: the standard base corresponding to the target sequencing round in the standard sample sequence is A, but the initial base recognition result of the target sequencing round base extension is not A. At this time, it is considered that the correct error type of the initial base recognition of the target sequencing round is: A base error category. In some embodiments, the A base error category can be further classified according to the initial base recognition result of the target sequencing round base extension, exemplarily, the category of A base error recognition as T, the category of A base error recognition as G, and the category of A base error recognition as C.

[0104] Similarly, the classification basis of the T or U base error category is: the standard base corresponding to the target sequencing round in the standard sample sequence is T, but the initial base recognition result of the target sequencing round base extension is not T. At this time, it is considered that the correct error type of the initial base recognition of the target sequencing round is: T or U base error category. In some embodiments, the T or U base error category can be further classified according to the initial base recognition result of the target sequencing round base extension, illustratively, T or U base error recognition is A category, T or U base error recognition is G category and T or U base error recognition is C category.

[0105] The classification basis of the G base error category is: the standard base corresponding to the target sequencing round in the standard sample sequence is G, but the initial base recognition result of the target sequencing round base extension is not G. At this time, it is considered that the correct error type of the initial base recognition of the target sequencing round is: G base error category. In some embodiments, the G base error category can be further classified according to the initial base recognition result of the target sequencing round base extension, illustratively, the category of G base error recognition as T, the category of G base error recognition as A, and the category of G base error recognition as C.

[0106] The classification basis of the C base error category is: the standard base corresponding to the target sequencing round in the standard sample sequence is C, but the initial base recognition result of the target sequencing round base extension is not C. At this time, it is considered that the correct error type of the initial base recognition of the target sequencing round is: C base error category. In some embodiments, the C base error category can be further classified according to the initial base recognition result of the target sequencing round base extension, illustratively, the category of C base error recognition as A, the category of C base error recognition as G, and the category of C base error recognition as T.

[0107] In the above steps (14) and (15), the second sequencing information of the target sequencing round determined when the base extension reaction is performed on the standard sample is used as the feature value, and the correct or incorrect category of the initial base recognition is used as the target value to construct the training sample. The pre-constructed correct or incorrect classification model of the base recognition result is trained based on the training sample, and the mapping relationship between the correct or incorrect classification information of the initial base recognition result of the sequencing round and the second sequencing information is obtained to obtain the trained correct or incorrect classification model of the base recognition result.

[0108] In one implementation scenario, the first initial base identification information includes the initial base identification information sequence of a consecutive sequencing rounds, that is, the context sequence of the target sequencing round (including the target sequencing round). In this case, the first initial base identification information is a sequence formed by a consecutive base identification information. When training the base identification result correct or incorrect classification model, the initial base identification information sequence of a consecutive sequencing rounds is taken as a whole and input into a pre-built machine learning model / deep learning model as a feature for training.

[0109] In another implementation, the first initial base identification information includes at least one of initial base identification information sequences of n1 sequencing rounds and initial base identification information sequences of m1 sequencing rounds, and initial base identification information of a target sequencing round.

[0110] In some embodiments, the first initial base identification information may be the initial base identification information sequence of n1 consecutive sequencing rounds and the initial base identification information of the target sequencing round. When training the base identification result correct or incorrect classification model, the initial base identification information sequence of n1 consecutive sequencing rounds is taken as a whole, together with the initial base identification information of the target sequencing round, and input into the pre-built machine learning model / deep learning model as two features for training.

[0111] In some embodiments, the first initial base identification information may be the initial base identification information sequence of m1 consecutive sequencing rounds and the initial base identification information of the target sequencing round. When training the base identification result correct or incorrect classification model, the initial base identification information sequence of m1 consecutive sequencing rounds is taken as a whole, together with the initial base identification information of the target sequencing round, and input into the pre-built machine learning model / deep learning model as two features for training.

[0112] In some embodiments, the first initial base identification information may be a collection of the initial base identification information sequences of n1 consecutive sequencing rounds, the initial base identification information sequences of m1 consecutive sequencing rounds, and the initial base identification information of the target sequencing round. When training the base identification result correct or incorrect classification model, the initial base identification information sequences of n1 consecutive sequencing rounds as a whole, the initial base identification information sequences of m1 consecutive sequencing rounds as a whole, together with the initial base identification information of the target sequencing round, are input into a pre-built machine learning model / deep learning model with three features for training.

[0113] In yet another implementation scenario, the first initial base identification information includes initial base identification information of each of a consecutive sequencing rounds, that is, a collection of initial base identification information of each of a consecutive sequencing rounds.

[0114] In some embodiments, the first initial base identification information is a set of initial base identification information of each sequencing round in n1 consecutive sequencing rounds and initial base identification information of the target sequencing round. When training the base identification result correct or incorrect classification model, the initial base identification information of n1+1 consecutive sequencing rounds is each input as a feature (a total of n1+1 features) into a pre-built machine learning model / deep learning model for training.

[0115] In some embodiments, the first initial base identification information is a set of initial base identification information of each sequencing round in m1 consecutive sequencing rounds and initial base identification information of the target sequencing round. When training the base identification result correct or incorrect classification model, the initial base identification information of each of the m1+1 consecutive sequencing rounds is each input as a feature (a total of m1+1 features) into a pre-built machine learning model / deep learning model for training.

[0116] In some embodiments, the first initial base identification information is a set of initial base identification information of each sequencing round in n1 consecutive sequencing rounds, initial base identification information of each sequencing round in m1 consecutive sequencing rounds, and initial base identification information of the target sequencing round. At this time, when training the correct and incorrect classification model of base identification results, the initial base identification information of each of the n1+m1+1 consecutive sequencing rounds is each input as a feature (a total of n1+m1+1 features) into the pre-built machine learning model / deep learning model for training.

[0117] In the embodiment of the present application, the mapping relationship between the correct and incorrect categories of the sequencing round and the second sequencing information can be determined by mathematical models, neural networks, etc., which are not limited here. It is sufficient to obtain the correct and incorrect categories of the initial base recognition of the known sample and the second sequencing information of the known sample, that is, to determine the correct and incorrect categories of the initial base recognition of the known sample according to the second sequencing information of the known sample.

[0118] S120: Determine correct or incorrect classification information of the initial base recognition result of the sequencing round to be processed based on the first mapping relationship and the first sequencing information of the sequencing round to be processed.

[0119] In this step, the sequencing round that completes the base extension reaction is used as the sequencing round to be processed, and the correct and incorrect classification information of the initial base recognition result of the sequencing round to be processed is determined according to the first sequencing information of the sequencing round to be processed and the first mapping relationship constructed in advance. At this time, the sequencing round to be processed is a sequencing round that needs to be base recognized, which is the sequencing round of the target, corresponding to the target sequencing round described above. Among them, the first sequencing information of the sequencing round to be processed includes the first initial base recognition information of a consecutive sequencing rounds, and a consecutive sequencing rounds include the sequencing round to be processed. The continuous a sequencing rounds and their first initial base recognition information can be referred to above, and will not be repeated here.

[0120] In one embodiment, determining the correctness or incorrectness classification information of the initial base recognition result of the sequencing round to be processed based on the first mapping relationship and the first sequencing information of the sequencing round to be processed includes:

[0121] S121, obtaining first sequencing information of a sequencing round to be processed;

[0122] S122: Determine correct or incorrect classification information of the initial base recognition result of the sequencing round to be processed based on the first sequencing information and the first mapping relationship.

[0123] In step S121, obtaining the first sequencing information of the sequencing round to be processed can be understood as collecting the first sequencing information of the sequencing round to be processed. The content covered by the first sequencing information is described above and will not be repeated here.

[0124] In step S122, in the method, a first mapping relationship between the standard base of the sequencing round and the first sequencing information is obtained by pre-training a machine learning model and / or a deep learning model between the standard base of the sequencing round and the first sequencing information, and then when performing base recognition on the sequencing round of the sample to be tested, the correct and incorrect classification information of the initial base recognition result of the sequencing round to be processed is determined by the first sequencing information of the sequencing round to be processed and the first mapping relationship. Specifically, the first sequencing information of the sequencing round to be processed is input into the first mapping relationship, and based on the association between the first sequencing information and the correct and incorrect classification information of the initial base recognition result of the sequencing round to be processed in the first mapping relationship, the correct and incorrect classification information of the initial base recognition result of the sequencing round to be processed is determined.

[0125] In some embodiments, the correct and incorrect classification information of the initial base recognition result is the probability score of the correct and incorrect category, or the probability value of the correct and incorrect category, that is, the percentage value corresponding to each category in the correct and incorrect category.

[0126] In some embodiments, the base recognition method may further include:

[0127] S130: Determine the base recognition result of the sequencing round to be processed based on the correct and incorrect classification information of the initial base recognition result.

[0128] As an implementation scenario, the correctness or error classification of the initial base recognition result is a correct recognition result, which is used as the base recognition result of the sequencing round to be processed. As another implementation scenario, when the correctness or error classification is the five classifications or more classifications as described above, the base recognition result of the sequencing round to be processed can also be determined based on the specific error category in the error classification. Exemplarily, when the correctness or error classification of the initial base recognition result is the A base error category, the A base is used as the base recognition result of the sequencing round to be processed; when the correctness or error classification of the initial base recognition result is the T or U base error category, the T or U base is used as the base recognition result of the sequencing round to be processed; when the correctness or error classification of the initial base recognition result is the G base error category, the G base is used as the base recognition result of the sequencing round to be processed; when the correctness or error classification of the initial base recognition result is the C base error category, the C base is used as the base recognition result of the sequencing round to be processed.

[0129] In some embodiments, the correct and incorrect classification information of the initial base recognition result is the probability score of the correct and incorrect category, or the probability value of the correct and incorrect category, that is, the percentage value corresponding to each category in the correct and incorrect category.

[0130] In some embodiments, step S130 includes:

[0131] S131, obtaining a second mapping relationship between a standard base of a sequencing round and the third sequencing information;

[0132] S132: Determine a base recognition result of the sequencing round to be processed based on the second mapping relationship and the third sequencing information of the sequencing round to be processed.

[0133] In the above step S131, the second mapping relationship between the standard base of the sequencing round and the third sequencing information refers to the relationship that the third sequencing information based on the sequencing round can correspond to the standard base or the real base of the sequencing round.

[0134] In the embodiments of the present application, the standard base of the sequencing round refers to the real base of the sequencing round. For the standard sample, the real base of the sequencing round is known, while for the unknown sample to be tested, the real base of the sequencing round is unknown. Therefore, the real base of the sequencing round is referred to as the standard base of the sequencing round.

[0135] In this embodiment of the present application, the third sequencing information of the sequencing round at least includes:

[0136] (a) The correct or incorrect classification information of the initial base recognition result obtained in step S120.

[0137] The correct and incorrect classification information of the initial base recognition results is as described above, and will not be repeated here to save space.

[0138] In some embodiments, the correctness and error classification information of the initial base recognition result is a binary correctness and error category, including a correct category and an incorrect category of the initial base recognition result. In other embodiments, the correctness and error classification information of the initial base recognition result is a multi-category with a classification number greater than 2. In one embodiment, the correctness and error category is five-category, and the five-category includes the following five categories:

[0139] The correct category of the initial base calling result;

[0140] A base error category;

[0141] T or U base error category;

[0142] G base error category;

[0143] C base error category.

[0144] The specific circumstances and explanations of the above two-category true and false categories, as well as the five-category true and false categories, are described above.

[0145] In some embodiments, the correct and incorrect classification information of the initial base recognition result is the probability score of the correct and incorrect category, or the probability value of the correct and incorrect category, that is, the percentage value corresponding to each category in the correct and incorrect category.

[0146] In addition, the third sequencing information also includes:

[0147] (b) Second initial base call information based on the target sequencing round.

[0148] In some embodiments, the second initial base recognition information includes initial base recognition information of the target sequencing round. It can also be said that the second initial base recognition information refers to the initial information obtained when performing base recognition on the target sequencing round. The initial information can be the real or objective raw data output by the base recognition, or it can be relative data obtained after processing the raw data output by the base recognition, but not as the final base recognition result.

[0149] In some embodiments, the second initial base identification information is the photoelectric signal intensity generated by the base extension reaction of the target sequencing round. In some embodiments, the photoelectric signal intensity includes the optical signal intensity of multiple optical signal channels, wherein the multiple optical signals correspond to the multiple base types involved in the base extension reaction. In this embodiment, there is a corresponding relationship between the optical signal and the base type and the optical signal channel. In one embodiment, each base type corresponds to an optical signal channel, and the signal generated by the base in the optical signal channel is collected to obtain a corresponding optical signal, such as an optical signal set of a specific wavelength. Exemplarily, based on the base type, the optical signal channel for collecting optical signals is four channels, including an A base channel, a C base channel, a G base channel, and a T / U base channel. In another embodiment, two or more base types obtain their respective optical signals through the same optical signal channel, and each signal in the optical signal channel is collected to obtain an optical signal corresponding to the number of base types, such as an optical signal set of two or more specific wavelengths, and each optical signal set contains all optical signals obtained at a specific wavelength. Exemplarily, based on the base type, the optical signal channel for collecting the optical signal is a dual channel, including an A / T / U base channel and a G / C base channel.

[0150] In some embodiments, the second initial base recognition information includes the optical signal intensity of each optical signal channel. Taking four channels as an example, the optical signal intensity from the A base channel, the C base channel, the G base channel and the T / U base channel is used as a component of the second initial base recognition information.

[0151] In some embodiments, the second initial base recognition information is the optical signal intensity with the largest intensity among the multiple optical signal channels. Taking four channels as an example, if the optical signal intensity in the A base channel is 0.9, and the optical signal intensities in the C base channel, the G base channel, and the T / U base channel are 0.2, 0.2, and 0.4, respectively, the optical signal intensity in the A base channel is used as a component of the second initial base recognition information.

[0152] In an embodiment in which a base fluorescence signal is used as a base recognition signal, the photoelectric signal intensity is the fluorescence signal intensity. At this time, the second initial base recognition information includes the fluorescence signal intensity of the fluorescence channel of the target sequencing round. Wherein, based on the base type, the fluorescence channel may include an A base fluorescence channel, a C base fluorescence channel, a G base fluorescence channel, and a T / U base fluorescence channel.

[0153] In an embodiment of the present application, the optical signal intensity includes any one of the initial optical signal intensity and the corrected optical signal intensity. Among them, the initial optical signal intensity refers to the optical signal intensity determined directly from the image, and the optical signal intensity can be obtained by extracting the intensity of the optical signal of the bright spot or the effective bright spot in the image. The corrected optical signal intensity refers to the intensity obtained after the initial optical signal intensity is corrected by a preset correction method. The preset correction method includes background correction, crosstalk correction, and reaction asynchrony correction, etc. In addition, the normalization process can also be understood as a kind of correction method, that is, the correction intensity includes at least one of the background correction intensity, the crosstalk correction intensity, the reaction asynchrony correction intensity, and the intensity after normalization, but is not limited to this.

[0154] In some embodiments using base fluorescence signals as base recognition signals, the optical signal intensity may be: the fluorescence intensity directly extracted from the optical image collected by the optical imaging device, i.e., the initial fluorescence intensity; or the image of the base collected by the optical imaging device is processed, such as the relative fluorescence intensity obtained after background correction. Of course, the optical signal intensity is not limited to the above-mentioned cases, and may also be other data obtained after processing the above-mentioned cases, such as the fluorescence intensity.

[0155] In one embodiment, the second initial base call information includes the corrected intensity obtained from the base extension reaction of the target sequencing round.

[0156] Exemplarily, the method for confirming the correction intensity includes: taking four channels as an example, collecting optical images obtained from the A base channel, C base channel, G base channel and T / U base channel in each sequencing round extension reaction, extracting the original optical signal intensity at each bright spot in the optical image; for each bright spot, the original optical signal intensity of the A base channel, C base channel, G base channel and T / U base channel is used as the initial intensity of the A base, C base, G base and T / U base respectively; performing at least one of background correction, crosstalk correction, reaction asynchrony correction and normalization processing on the initial intensity, and the correction intensity of the A base, C base, G base and T / U base can be determined. In this embodiment, the intensity after correction is used as the source of the first initial base recognition information, which can remove interference caused by factors such as background, crosstalk and reaction asynchrony, and improve the purity of the sequencing signal; when it is used as a feature value to input into the corresponding machine learning and / or deep learning model for machine learning and / or deep learning, it is more conducive to improving the accuracy of model training.

[0157] In some embodiments, the third sequencing information further includes at least one of the following information:

[0158] (c) E value from the target sequencing round;

[0159] (d) the sequence determined by the E value of each sequencing round in b consecutive sequencing rounds;

[0160] (e) Photoelectric signal intensity generated by each sequencing round in b consecutive sequencing rounds.

[0161] For (c) E value from the target sequencing round, the E value is used to evaluate the quality of sequencing data and can usually characterize the accuracy of sequencing data. Specifically, the E value is the ratio of the maximum value of the photoelectric signal intensity of the four base types in the optical detection channel to the sum of the photoelectric signal intensities of the four base types. Exemplarily, in a sequencing method for collecting fluorescence signals through four channels, for a bright spot corresponding to a nucleic acid template, the E value is the ratio of the photoelectric signal intensity in the channel with the strongest photoelectric signal to the photoelectric signal intensity of the four channels.

[0162] For (d) the sequence determined by the E value of each sequencing round in b consecutive sequencing rounds, it refers to the E value sequence formed by arranging the E value of each sequencing round in the order of the sequencing rounds in b consecutive sequencing rounds, wherein the E value is the ratio of the maximum value of the photoelectric signal intensity of the four base types in the optical detection channel to the sum of the photoelectric signal intensities of the four base types. As a sequence as a whole, it is associated with the standard base of the target sequencing round. It should be understood that during machine learning and / or deep learning, the sequence determined by the E value of each sequencing round in b consecutive sequencing rounds is used as a feature for training the base recognition model.

[0163] In some embodiments, the b consecutive sequencing rounds include a target sequencing round, and at least one of the following two situations:

[0164] n2 sequencing rounds that perform base extension reactions prior to the target sequencing round,

[0165] m2 sequencing rounds in which base extension reaction is performed later than the target sequencing round.

[0166] That is, the consecutive b sequencing rounds include three implementation scenarios: in the first implementation scenario, the consecutive b sequencing rounds include the target sequencing round, and n2 sequencing rounds in which base extension reactions occur before the target sequencing round, i.e., the first n2 sequencing rounds, and in this case, b=n2+1; in the second implementation scenario, the consecutive b sequencing rounds include the target sequencing round, and m2 sequencing rounds in which base extension reactions occur after the target sequencing round, i.e., the last m2 sequencing rounds, and in this case, b=m2+1; in the third implementation scenario, the consecutive b sequencing rounds include the target sequencing round, n2 sequencing rounds in which base extension reactions occur before the target sequencing round, i.e., the first n2 sequencing rounds, and m2 sequencing rounds in which base extension reactions occur after the target sequencing round, i.e., the last m2 sequencing rounds, and in this case, b=n2+m2+1, and b is a natural number greater than or equal to 2.

[0167] Among them, the values ​​of m2 and n2 satisfy: n2 is an integer greater than or equal to 0, and m2 is an integer greater than or equal to 0. When m2 and n2 are both 0, the consecutive b sequencing rounds are the target sequencing rounds, that is, the sequencing rounds that require base recognition, which repeats with (b). Therefore, in the embodiment of the present application, m2 and n2 are not 0 at the same time, that is: at least one of m2 and n2 is not 0. Exemplarily, at least one of m2 and n2 is not 0, including: m2 is 0, n2 is not 0; m2 is not 0, n2 is 0; m2 is not 0, and n2 is not 0.

[0168] In some embodiments, m2 and n2 are each a natural number from 1 to 50. For example, m2 can be 1, 2, 3, 4, 5, 8, 10, 12, 15, 18, 20, 22, 25, 28, 30, 32, 35, 38, 40, 42, 45, 48, 50, etc., and n2 can be 1, 2, 3, 4, 5, 8, 10, 12, 15, 18, 20, 22, 25, 28, 30, 32, 35, 38, 40, 42, 45, 48, 50, etc. It should be understood that the above values ​​are only examples. In fact, the values ​​of m2 and n2 are not limited thereto.

[0169] It should be understood that when the b consecutive sequencing rounds simultaneously include n2 sequencing rounds in which base extension reactions occur before the target sequencing round, i.e., the first n2 sequencing rounds, and m2 sequencing rounds in which base extension reactions occur after the target sequencing round, i.e., the last m2 sequencing rounds, there is no strict requirement for the selection of m2 and n2, and they can be the same or different.

[0170] In some embodiments, the third sequencing information includes both (c) and (d) above, that is, the third sequencing information includes both the E value from the target sequencing round and the sequence determined by the E value from each sequencing round in b consecutive sequencing rounds.

[0171] (e) The photoelectric signal intensity generated by each sequencing round in b consecutive sequencing rounds refers to the photoelectric signal intensity obtained by performing base recognition in b consecutive sequencing rounds. The description of the photoelectric signal intensity in b consecutive sequencing rounds is described above and will not be repeated here.

[0172] In some embodiments, the photoelectric signal intensity generated by each sequencing round in b consecutive sequencing rounds, in addition to the photoelectric signal intensity generated by the target sequencing round, also includes the photoelectric signal intensity generated by the first n2 sequencing rounds and / or the photoelectric signal intensity generated by the last m2 sequencing rounds.

[0173] In the embodiment of the present application, the second mapping relationship between the standard base of the sequencing round and the third sequencing information can be a mapping mathematical model, a mapping neural network model, etc. The second mapping relationship can be obtained by taking the third sequencing information of the target sequencing round in the standard sample as a feature value and the standard base of the standard sample as a target value, and inputting them into a pre-set machine learning and / or deep learning model for machine learning and / or deep learning, that is, the mapping relationship between the standard base of the sequencing round and the third sequencing information is obtained by training a pre-constructed base recognition model.

[0174] In some embodiments, training of the base recognition model comprises:

[0175] (21) obtaining a standard sample and a sequence of the standard sample;

[0176] (22) performing a base extension reaction on the standard sample to determine the third sequencing information of the sequencing round;

[0177] (23) constructing training samples using the third sequencing information of each sequencing round as a feature value and the standard base corresponding to the sequencing round as a target value;

[0178] (24) The pre-constructed base recognition model is trained based on the training samples to obtain a trained base recognition model.

[0179] In the above step (21), the standard sample refers to a nucleic acid sample with a known base sequence. In the embodiment of the present application, there is no strict requirement for the selection of the standard sample. It can be a human reference gene sequence or a reference gene sequence of other species. It only needs to be a known sequence. In some embodiments, a sample with a larger genome is selected as the standard sample; samples containing multiple different species can also be used as standard samples together. Thus, by deriving the training data from a larger genome sample or a genome sample of a non-single species, the training data can cover more features, especially the features of the context sequence required by the embodiment of the present application, thereby making the base recognition model more generalizable.

[0180] The sequence of the standard sample can be obtained directly from the information provided by the standard sample, or by sequencing the standard sample and comparing it with the reference genome to determine its sequence. Exemplarily, on a selected sequencing platform, the standard sample is sequenced and the corresponding base sequence is obtained; the obtained base sequence is compared with the reference genome of the standard sample to obtain a known sequence.

[0181] In the above step (22), a base extension reaction is performed on the standard sample to obtain the third sequencing information of the sequencing round. In the embodiment of the present application, the third sequencing information is as described above and will not be repeated here.

[0182] In the above steps (23) and (24), the third sequencing information of each sequencing round is used as a feature value, and the standard base corresponding to the sequencing round is used as a target value to construct a training sample. The pre-constructed base recognition model is trained based on the training sample to obtain the mapping relationship between the third sequencing information of the sequencing round and the standard base, and obtain the trained base recognition model.

[0183] It should be understood that when training the base recognition result correctness or incorrectness classification model, when the third sequencing information contains sequence features, the sequence as a whole is input into the pre-built machine learning model / deep learning model as one feature for training.

[0184] In the above step S132, based on the second mapping relationship and the third sequencing information of the sequencing round to be processed, the base recognition result of the sequencing round to be processed can be determined. In some embodiments, the output form of the base recognition result can be a base type, such as A base, T or U base, G base and C base; it can also be the probability score of each base type being detected or recognized, exemplarily, the probability score of A base is 94%, the probability score of T or U base is 2%, the probability score of G base is 2% and the probability score of C base is 2%. It should be understood that in conventional base recognition software, as a general understanding, if not otherwise specified, the base type of the sequencing round output by base recognition can be determined by the probability of each base type. Specifically, the base with the highest probability among the four bases is considered to be the base type obtained by base recognition.

[0185] In one embodiment, in step S132, determining the initial base recognition result of the sequencing round to be processed based on the second mapping relationship and the third sequencing information of the sequencing round to be processed includes:

[0186] Obtaining the third sequencing information of the sequencing round to be processed;

[0187] Based on the third sequencing information and the second mapping relationship, the probabilities of the four base types of the sequencing round to be processed are determined, and the base type with the greatest probability is used as the base recognition result of the sequencing round to be processed.

[0188] So far, the base recognition method provided in this embodiment involves two machine learning / deep learning models. The first base recognition correct and incorrect classification model model is trained and learned separately by separating the correct and incorrect influence of the sequence context on the target sequencing round, and then the model result is used for the training of the base recognition model, that is, first establish a base recognition correct and incorrect classification model based on context sequence information, and then establish a base recognition model, so that the feature space dimension of the base recognition model can be reduced, and the dimensionality disaster can be avoided, thereby avoiding the demand for a large amount of training data and training computing power, reducing the risk of overfitting, and improving the stability and generalization of the model. Specifically, the base recognition correct and incorrect classification model requires more training data, and although the base recognition model requires relatively less training data, the feature parameter information that needs to be extracted for each training data is relatively more. Although more training data is required for training the base recognition correct and incorrect classification model, due to the small number of feature values ​​(such as: the feature value of the base recognition correct and incorrect classification model can be only one: the context is merged as a feature, or two: the above is a feature and the following is a feature), the memory pressure is relatively small during storage and training. The features required for training the base recognition model include the fluorescence brightness of the four channels corresponding to the current base (4 features), and may include the current E value, the fluorescence brightness of the adjacent front and rear wheels (8 features in total), etc., and the features are significantly increased. Compared with extracting and training the features required by the two models together, this embodiment uses two models to train independently, and does not require the use of exactly the same training data set and training data volume, nor does it require a large number of training data entries, which has advantages in terms of computing power, storage, training time, and model stability.

[0189] In a possible embodiment, the base recognition method further comprises:

[0190] S133. Based on the base recognition result, determine the base recognition error rate of the sequencing round to be processed.

[0191] In this embodiment, the base recognition result is the probability score of each base type being detected or recognized, that is, the recognition probability scores of the four base types. The probability scores of each base type being detected or recognized are combined to determine the base recognition error rate of the sequencing round to be processed.

[0192] In some embodiments, step S133 determines the base recognition error rate of the sequencing round to be processed based on the base recognition result, including:

[0193] S1331, obtaining a third mapping relationship between fourth sequencing information of a sequencing round and a base recognition error rate;

[0194] S1332: Determine the base recognition error rate of the sequencing round to be processed based on the third mapping relationship and the fourth sequencing information of the sequencing round to be processed.

[0195] In the above step S1331, the third mapping relationship between the base recognition error rate of the sequencing round and the fourth sequencing information refers to the relationship that the fourth sequencing information based on the sequencing round can correspond to the base recognition error rate of the sequencing round.

[0196] In the embodiments of the present application, the base recognition error rate is used to characterize the accuracy of the base recognition results of the sequencing round. In some embodiments, the base recognition error rate can characterize the error probability of four base types, which can be characterized by the Q value, and the calculation formula of the Q value is as follows:

[0197] Q = -10*log10(P)

[0198] Where Q is the quality score and P is the probability of base calling error.

[0199] In the embodiment of the present application, the fourth sequencing information of the sequencing round is information associated with the features generated by the base extension reaction of the sequencing round. In some embodiments, the fourth sequencing information includes at least four base type recognition probability scores. That is, the fourth sequencing information includes the initial base recognition result determined in step S132, and the initial base recognition result is the probability score of each base type being detected or recognized, that is, the recognition probability scores of the four base types.

[0200] In some embodiments, the fourth sequencing information further includes the correctness classification information of the initial base recognition result obtained in step S120. In one implementation, the correctness classification information of the initial base recognition result is a five-category correctness classification, and the five-category classification includes the following five categories:

[0201] The correct category of the initial base calling result;

[0202] A base error category;

[0203] T or U base error category;

[0204] G base error category;

[0205] C base error category.

[0206] The specific circumstances and explanations of the five categories of right and wrong are described above and will not be repeated here to save space.

[0207] In the embodiment of the present application, the third mapping relationship between the fourth sequencing information of the sequencing round and the base recognition error rate can be a mapping mathematical model, a mapping neural network model, etc. The third mapping relationship can be obtained by taking the fourth sequencing information of the target sequencing round in the standard sample as a feature value and the base recognition error rate of the standard sample as a target value, and inputting them into a pre-set machine learning and / or deep learning model for machine learning and / or deep learning, that is, the mapping relationship between the base recognition error rate of the sequencing round and the fourth sequencing information is obtained by training the pre-constructed base recognition error rate model.

[0208] In some embodiments, training the base calling error rate model includes:

[0209] (31) obtaining a standard sample and a sequence of the standard sample;

[0210] (32) performing a base extension reaction on the standard sample to determine the fourth sequencing information of the sequencing round;

[0211] (33) constructing training samples using the fourth sequencing information of each sequencing round as a feature value and the base recognition error rate corresponding to the sequencing round as a target value;

[0212] (34) A pre-constructed base recognition error rate model is trained based on the training samples to obtain a trained base recognition error rate model.

[0213] In the above step (31), the standard sample refers to a nucleic acid sample with a known base sequence. In the embodiment of the present application, there is no strict requirement for the selection of the standard sample. It can be a human reference gene sequence or a reference gene sequence of other species. It only needs to be a known sequence. In some embodiments, a sample with a larger genome is selected as the standard sample; samples containing multiple different species can also be used as standard samples together. Thus, by deriving the training data from a larger genome sample or a genome sample of a non-single species, the training data can cover more features, especially the features of the context sequence required by the embodiment of the present application, thereby making the base recognition model more generalizable.

[0214] The sequence of the standard sample can be obtained directly from the information provided by the standard sample, or by sequencing the standard sample and comparing it with the reference genome to determine its sequence. Exemplarily, on a selected sequencing platform, the standard sample is sequenced and the corresponding base sequence is obtained; the obtained base sequence is compared with the reference genome of the standard sample to obtain a known sequence.

[0215] In the above step (32), a base extension reaction is performed on the standard sample to obtain the fourth sequencing information of the sequencing round. In the embodiment of the present application, the fourth sequencing information includes at least the recognition probability scores of four base types. Optionally, it also includes the correct and incorrect classification information of the initial base recognition result, such as five classifications. The fourth sequencing information refers to the above, and the fourth sequencing information is not described in detail here.

[0216] In the above steps (33) and (34), the fourth sequencing information of each sequencing round is used as a feature value, and the base recognition error rate corresponding to the sequencing round is used as a target value to construct a training sample. The pre-constructed base recognition error rate model is trained based on the training sample to obtain a mapping relationship between the fourth sequencing information of the sequencing round and the base recognition error rate, and obtain a trained base recognition error rate model.

[0217] It should be understood that when training the base recognition error rate model, the recognition probability scores of the four base types are each used as a feature, that is, the four features are respectively input into the pre-built machine learning model / deep learning model for training. When the fourth sequencing information contains the correct and incorrect classification information of five categories, the five categories are each used as a feature, that is, five features are respectively input into the pre-built machine learning model / deep learning model for training, plus the recognition probability scores of the four base types. At this time, at least nine features are respectively input into the pre-built machine learning model / deep learning model for training.

[0218] In the above step S1332, based on the third mapping relationship and the fourth sequencing information of the sequencing round to be processed, the base recognition error rate of the sequencing round to be processed can be determined. In one embodiment, in step S1332, based on the third mapping relationship and the fourth sequencing information of the sequencing round to be processed, the base recognition error rate of the sequencing round to be processed is determined, including:

[0219] Obtaining fourth sequencing information of the sequencing round to be processed;

[0220] Based on the fourth sequencing information and the third mapping relationship, a base recognition error rate of the sequencing round to be processed is determined.

[0221] In some embodiments, further, the base recognition error rate is converted into a Q value. In some embodiments, the base recognition error rate is converted into a Q value, which can be converted by the following calculation formula: Q = -10*log10(P). Wherein, Q is the quality score, and P is the probability of base recognition error.

[0222] The base recognition method provided by the embodiment of the present application can add the used model as a supplementary module to the original base recognition algorithm software of the sequencing platform. In the sequencing process, firstly, the original algorithm software is used to perform normal base recognition of sequencing according to the original time node, and then the method used by the embodiment of the present application is used to re-identify or correct the base recognition of the target sequencing round. Of course, since the method needs to use context information, when the original base recognition algorithm software of the sequencing platform is sequenced and identified for the Nth round of bases, it is likely that the following sequence cannot be used temporarily to identify the base of the target sequencing round, and it is necessary to wait until the original base recognition algorithm software completes base recognition for the predetermined number of following rounds after the Nth round, and then the following sequence information is used to re-identify and correct the base of the Nth round.

[0223] The base recognition method provided in this embodiment is described below through specific examples.

[0224] The base recognition method provided in this embodiment includes correcting the base recognition results based on a consecutive sequencing rounds, including:

[0225] (1) A whole genome (WGS) library was constructed using the human genome standard sample HG001, and 150 rounds of double-end sequencing of a standard single chip were completed on the sequencing platform, and base recognition was completed using the existing supporting base recognition software. The software can locate the position of the DNA template cluster to be tested on the original sequencing image, and then identify the bases corresponding to each round of sequencing for each template. The sequencing results were compared with the human genome reference genome HG19. It should be noted that data sets can also be constructed in combination with other genomes, such as the genomes of other species, to increase the diversity of the data.

[0226] (2) The sequencing data of 1200 different fields of view corresponding to 10 groups of experiments were used as the total data set, wherein the coverage bases corresponding to the polymorphic sites were removed from the total data set to avoid the situation where the reference genome is inaccurate and thus affects the model training (in this embodiment, the region of suspected polymorphic sites can be removed by purchasing standard products, referring to standard gene library information, etc.). The training set and test set of this embodiment are both sampled from the total data set, specifically:

[0227] (21) Extract training set 1 from the total data set.

[0228] Extract the context sequence and sequencing error information of each base: extract 30 sequencing bases of the adjacent context of the corresponding base as the context sequence, that is, the previous sequence and the following sequence are a string of 30 bases respectively. If the context sequence is less than 30 bases, use N to fill the context sequence away from the base to be analyzed to a total length of 30 bases. The error information of the base is set to five categories, namely "correct", "A base error (should be A)", "C base error (should be C)", "G base error (should be G)", "T or U base error (should be T or U)".

[0229] (22) Using the data from training set 1, the previous sequence, the following sequence, and the current actual base type read (the base type of the target sequencing round) corresponding to each base are used as input features, and the correct or incorrect information of the base is used as the target value. A classification model is established using a convolutional neural network to predict the impact of the context on the current base recognition. This model is referred to as the "base recognition correct or incorrect classification model". The base recognition correct or incorrect classification model outputs five probability scores corresponding to the five categories.

[0230] The feature value of the "context sequence" can be represented by any of the two methods in the above examples. Optionally, the context information can be input into the training by taking each base as an independent feature, or the entire context or the sequence of the previous context and the next context as a string (i.e., a feature). When the entire context is taken as a string as the input feature, the number of features of the model can be greatly reduced, reducing the complexity of the model.

[0231] (23) Extract training set 2 (training set 2 has no intersection with training set 1).

[0232] Extract the corrected brightness of the four fluorescence channels corresponding to each base, as well as the corrected fluorescence brightness of the four fluorescence channels corresponding to the previous and next bases. At the same time, extract the base quality score corresponding to the current base. Extract the base "answer" (standard base) on the reference genome after alignment corresponding to each base.

[0233] (24) Using the training set 2 data, the four-channel brightness corresponding to the current base, the previous base, and the next base, as well as the base quality score corresponding to the current base, and the five probability scores obtained after each base was classified using the base recognition correct or incorrect classification model established in the previous step (22) were used as input features, and the "answer" on the reference genome after alignment corresponding to each base was used as the target value. A machine learning model was established using Light GBM. This model is referred to as the "base recognition model" here.

[0234] It should be noted that when the base recognition model is established, the brightness information corresponding to a specific number of rounds can be selected as the feature input, and the context range of the brightness feature input and the context range of the sequence input do not need to be consistent. Therefore, a larger range of context sequences can be selected as input, while a smaller range of brightness information or even no context information can be selected as input.

[0235] (25) A test set is extracted from the total data set. This test set has no intersection with training set 1 and training set 2. On the test set, the total error rate drops by about 49%.

[0236] (3) Integrate the above-mentioned “base recognition correct and incorrect classification model” and “base recognition model” into the base recognition software supporting the above-mentioned sequencing platform to generate a new version of the base recognition software.

[0237] After the new version of the base recognition software is generated, the new version of the base recognition software is tested. Specifically, the human genome standard sample HG002 is used to construct a whole exome (WES) library. The GenoLab M platform is used with the new version of the base recognition software to perform standard single chip, double-end 150 rounds of sequencing. After the sequencing is completed, the original sequencing images of this experiment are analyzed again with the old version of the base recognition software to obtain the sequencing results under the old version of the base recognition software.

[0238] (4) Further bioinformatics analysis was performed on the sequencing results generated by the new and old versions of the base recognition software. Only the high confidence intervals of the HG002 library genome were selected as the analysis objects (based on the analysis of sequencing data from multiple platforms available on the Internet, the intervals with a probability greater than the preset probability that do not contain polymorphic sites and other difficult-to-process areas were screened out). For the above-mentioned "analysis objects", it is believed that the sequencing results covering the corresponding areas should be exactly the same as the bases in the corresponding areas of the human reference genome HG19, otherwise there is an error in the sequencing.

[0239] After the bioinformatics features of the sequencing results corresponding to the base recognition software before and after the improvement were counted separately. On the one hand, the sequencing error rate corresponding to the base recognition software of the present embodiment is 45% lower than the base recognition software before the improvement (the original base recognition software of the sequencing platform). On the other hand, the mutation frequency of each site in the above-mentioned "analysis object" area is counted. If the error rate of the base covered on a certain site exceeds 2% (positive and negative chains are analyzed separately), it is considered that this site is a "false mutation" site. Relative to the base recognition software before the improvement, the number of "false mutation" sites of the base recognition software of the present embodiment has been reduced by 79%. It can be seen from this that after using the base recognition software of the present embodiment, not only the overall error rate has decreased, but also the number of sites prone to high-frequency errors has also decreased significantly.

[0240] Figure 2 It is a base recognition device provided by an embodiment of the present invention, and the device is used to implement the steps of the above-mentioned base recognition method. The description of the technical features and advantages in the base recognition method implementation is also applicable to the device, and will not be repeated here. It can be understood that the additional technical features of the base recognition method in any of the above-mentioned embodiments, including sub-steps, additional steps, optional alternative or preferred settings or processing, etc., can be implemented by making the device or the module of the device further include a unit / module or a sub-unit / sub-module.

[0241] like Figure 2 As shown, the base recognition device comprises:

[0242] A first mapping relationship acquisition module 210 is used to acquire a first mapping relationship between the correct and incorrect classification information of the initial base recognition result of the sequencing round and the first sequencing information, wherein the first sequencing information includes the first initial base recognition information based on the target sequencing round, and the first initial base recognition information includes the initial base recognition information of a consecutive sequencing rounds, and the a consecutive sequencing rounds at least include the target sequencing round;

[0243] The right-or-wrong classification information determination module 220 is used to determine the right-or-wrong classification information of the initial base recognition result of the sequencing round to be processed based on the first mapping relationship and the first sequencing information of the sequencing round to be processed.

[0244] In some embodiments of the first mapping relationship acquisition module 210, the first initial base identification information includes initial base identification information of a consecutive sequencing rounds, and the a consecutive sequencing rounds at least include the designated sequencing round.

[0245] In some embodiments of the first mapping relationship acquisition module 210, the consecutive a sequencing rounds include a target sequencing round, and at least one of the following two situations:

[0246] n1 sequencing rounds that perform base extension reactions prior to the target sequencing round,

[0247] m1 sequencing rounds that perform base extension reactions later than the target sequencing round,

[0248] Wherein, n1 and m1 are each independently an integer greater than or equal to 0, and m1 and n1 are not 0 at the same time.

[0249] In some embodiments of the first mapping relationship acquisition module 210, the first initial base identification information includes a sequence of initial base identification information of consecutive a sequencing rounds.

[0250] In some embodiments of the first mapping relationship acquisition module 210, the first initial base recognition information further includes initial base recognition information of a target sequencing round.

[0251] In some embodiments of the first mapping relationship acquisition module 210, the first initial base identification information includes initial base identification information of each of the n1 sequencing rounds and / or the m1 sequencing rounds, and initial base identification information of the target sequencing round.

[0252] In some embodiments of the first mapping relationship acquisition module 210, the first initial base identification information includes initial base identification information of each sequencing round in the a consecutive sequencing rounds.

[0253] In some embodiments of the first mapping relationship acquisition module 210, the first initial base identification information includes an initial base identification information sequence formed by at least one of n1 sequencing rounds and / or m1 sequencing rounds and a target sequencing round, and the initial base identification information of the target sequencing round.

[0254] In some embodiments of the first mapping relationship acquisition module 210, the first initial base recognition information is the intensity of a photoelectric signal generated during a base extension reaction.

[0255] In some embodiments of the first mapping relationship acquisition module 210, the photoelectric signal intensity includes optical signal intensities of multiple optical signal channels, wherein the multiple optical signals respectively correspond to multiple base types participating in the base extension reaction.

[0256] In some embodiments of the first mapping relationship acquisition module 210 , the optical signal intensity is the optical signal intensity with the largest intensity among the multiple optical signal channels.

[0257] In some embodiments of the first mapping relationship acquisition module 210 , the optical signal intensity is the fluorescence signal intensity.

[0258] In some embodiments of the first mapping relationship acquisition module 210 , the fluorescence signal intensity includes any one of an initial fluorescence signal intensity and a corrected fluorescence signal intensity.

[0259] In some embodiments of the first mapping relationship acquisition module 210, the correction includes at least one of background correction, crosstalk correction, reaction asynchrony correction and normalization processing.

[0260] In some embodiments, the correct and incorrect classification information can be a correct and incorrect category, including a correct category and an incorrect category. In some embodiments, the correct and incorrect category can be a binary classification, i.e., a correct category and an incorrect category. In one implementation scenario, if the initial base recognition result of the sequencing round base extension is consistent with the true base alignment, the initial base recognition of the sequencing round is considered to be correct and is classified as the correct category in the correct and incorrect classification information. Exemplarily, the initial base recognition result of the sequencing round base extension is A, and the reagent base type is also A, then the initial base recognition of the sequencing round is considered to be correct, and at this time, it is classified as the correct category in the correct and incorrect classification information. In another implementation scenario, if the initial base recognition result of the sequencing round base extension is inconsistent with the true base alignment, then the initial base recognition of the sequencing round is considered to be wrong, and it is classified as the wrong category in the correct and incorrect classification information. Exemplarily, the initial base recognition result of the sequencing round base extension is A, and the true base is T, then the initial base recognition of the sequencing round is considered to be wrong, and at this time, it is classified as the wrong category in the correct and incorrect classification information.

[0261] In some embodiments, the error categories may be further classified according to different types of errors, so that the right and wrong categories are multiple categories with the number of categories being greater than 2. In one embodiment, the right and wrong categories are five categories, and the five categories include the following five categories:

[0262] The correct category of the initial base calling result;

[0263] A base error category;

[0264] T or U base error category;

[0265] G base error category;

[0266] C base error category.

[0267] The classification basis of the A base error category is: the actual base corresponding to the sequencing round is A, but the initial base recognition result of the sequencing round base extension is not A. At this time, it is considered that the correct error type of the initial base recognition of the sequencing round is: A base error category. In some embodiments, the A base error category can be further classified according to the initial base recognition result of the sequencing round base extension, illustratively, the category of A base error recognition as T, the category of A base error recognition as G, and the category of A base error recognition as C.

[0268] Similarly, the classification basis of the T or U base error category is: the actual base corresponding to the sequencing round is T, but the initial base recognition result of the sequencing round base extension is not T. At this time, it is considered that the correct error type of the initial base recognition of the sequencing round is: T or U base error category. In some embodiments, the T or U base error category can be further classified according to the initial base recognition result of the sequencing round base extension, illustratively, the T or U base error is recognized as the category of A, the T or U base error is recognized as the category of G, and the T or U base error is recognized as the category of C.

[0269] The classification basis of the G base error category is: the actual base corresponding to the sequencing round is G, but the initial base recognition result of the sequencing round base extension is not G. At this time, it is considered that the correct error type of the initial base recognition of the sequencing round is: G base error category. In some embodiments, the G base error category can be further classified according to the initial base recognition result of the sequencing round base extension, illustratively, the category of G base error recognition as T, the category of G base error recognition as A, and the category of G base error recognition as C.

[0270] The classification basis of the C base error category is: the actual base corresponding to the sequencing round is C, but the initial base recognition result of the sequencing round base extension is not C. At this time, it is considered that the correct error type of the initial base recognition of the sequencing round is: C base error category. In some embodiments, the C base error category can be further classified according to the initial base recognition result of the sequencing round base extension, illustratively, the category of C base error recognition as A, the category of C base error recognition as G, and the category of C base error recognition as T.

[0271] In some embodiments of the first mapping relationship acquisition module 210, the first mapping relationship is obtained by training a pre-constructed base recognition result correct and incorrect classification model.

[0272] In some embodiments, the training of the base recognition result correct and incorrect classification model includes:

[0273] Obtaining a standard sample and a sequence of the standard sample;

[0274] Performing a base extension reaction on the standard sample to determine second sequencing information of a target sequencing round and an initial base recognition result of the base extension of the target sequencing round, wherein the second sequencing information at least includes the first sequencing information;

[0275] Determining the correct or incorrect category of the initial base recognition of the target sequencing round based on the initial base recognition result of the target sequencing round base extension and the standard base;

[0276] The second sequencing information is used as the feature value and the correct or incorrect category of the initial base recognition sequence is used as the target value to construct a training sample;

[0277] The pre-constructed base recognition result correct / error classification model is trained based on the training samples to obtain a trained base recognition result correct / error classification model.

[0278] In some embodiments, the correct or incorrect classification information determination module 220 includes:

[0279] A first sequencing information submodule, used to obtain first sequencing information of a sequencing round to be processed;

[0280] The right-or-wrong classification information determination submodule is used to determine the right-or-wrong classification information of the initial base recognition result of the sequencing round to be processed based on the first sequencing information and the first mapping relationship.

[0281] In some embodiments, the base recognition device further comprises:

[0282] The base recognition result determination module 230 is used to determine the base recognition result of the sequencing round to be processed based on the correct and incorrect classification information of the initial base recognition result.

[0283] In some embodiments, the base call result determination module 230 includes:

[0284] A second mapping relationship acquisition module, used to acquire a second mapping relationship between a standard base of a sequencing round and third sequencing information, wherein the third sequencing information includes second initial base recognition information based on a target sequencing round and correct or incorrect classification information of the initial base recognition result;

[0285] The base recognition result determination submodule is used to determine the base recognition result of the sequencing round to be processed based on the second mapping relationship and the third sequencing information of the sequencing round to be processed.

[0286] In some embodiments, the base recognition result determination submodule is specifically used to:

[0287] Obtaining the third sequencing information of the sequencing round to be processed;

[0288] Based on the third sequencing information and the second mapping relationship, the probabilities of the four base types of the sequencing round to be processed are determined, and the base type with the greatest probability is used as the base recognition result of the sequencing round to be processed.

[0289] In some embodiments, the second initial base call information includes initial base call information of a target sequencing round;

[0290] The second initial base recognition information is the photoelectric signal intensity generated when the target sequencing round performs a base extension reaction.

[0291] In some embodiments, the photoelectric signal intensity includes optical signal intensities of multiple optical signal channels, wherein the multiple optical signals respectively correspond to multiple base types participating in the base extension reaction.

[0292] In some embodiments, the optical signal intensity is the intensity of the optical signal with the highest intensity among the multiple optical signal channels.

[0293] In some embodiments, the optical signal intensity is a fluorescent signal intensity.

[0294] In some embodiments, the fluorescence signal intensity includes any one of an initial fluorescence signal intensity and a corrected fluorescence signal intensity.

[0295] In some embodiments, the correction includes at least one of background correction, crosstalk correction, reaction asynchrony correction, and normalization processing.

[0296] In some embodiments, the third sequencing information further includes at least one of the following information:

[0297] The E value from the target sequencing round;

[0298] The sequence determined by the E value of each sequencing round in b consecutive sequencing rounds;

[0299] The photoelectric signal intensity generated by each sequencing round in b consecutive sequencing rounds,

[0300] The E value is used to evaluate the quality of sequencing data and can usually characterize the accuracy of sequencing data. Specifically, the E value is the ratio of the maximum value of the photoelectric signal intensity of the four base types in the optical detection channel to the sum of the photoelectric signal intensities of the four base types.

[0301] In some embodiments, the b consecutive sequencing rounds are selected from at least one of the following:

[0302] Including a target sequencing round and n2 sequencing rounds that perform base extension reactions prior to the target sequencing round; or

[0303] Including a target sequencing round and m2 sequencing rounds that perform a base extension reaction later than the target sequencing round; or

[0304] Including a target sequencing round, n2 sequencing rounds that complete sequencing before the target sequencing round, and m2 sequencing rounds that complete sequencing after the target sequencing round;

[0305] Wherein, n2 and m2 are each independently an integer greater than or equal to 0, and m2 and n2 are not 0 at the same time.

[0306] In some embodiments, the base recognition device further comprises:

[0307] The base recognition error rate determination module 240 is used to determine the base recognition error rate of the sequencing round to be processed based on the base recognition result.

[0308] In some embodiments, the base calling error rate determination module includes:

[0309] A third mapping relationship acquisition module, used to acquire a third mapping relationship between fourth sequencing information of a sequencing round and the base recognition error rate, wherein the fourth sequencing information includes recognition probability scores of four base types;

[0310] The base recognition error rate determination submodule is used to determine the base recognition error rate of the sequencing round to be processed based on the third mapping relationship and the fourth sequencing information of the sequencing round to be processed.

[0311] In some embodiments, the fourth sequencing information further includes correctness or error classification information of the initial base recognition result, and the correctness or error classification information includes the following five categories:

[0312] The correct category of the initial base calling result;

[0313] A base error category;

[0314] T or U base error category;

[0315] G base error category;

[0316] C base error category.

[0317] The base recognition device provided in the embodiment of the present invention can execute the base recognition method provided in any embodiment of the present invention, and has a module corresponding to the execution method.

[0318] It should be understood that the features involved in the base recognition method can all be used in the base recognition device to explain the base recognition device and each module to implement the steps of the base recognition method.

[0319] Figure 3 1 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. The electronic device 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0320] like Figure 3As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0321] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0322] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 11 performs the various methods and processes described above, such as base recognition methods.

[0323] In some embodiments, the base identification method can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the base identification method described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to perform the base identification method in any other appropriate manner (for example, by means of firmware).

[0324] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0325] The computer program for implementing the base identification method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to the processor of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the computer program, when executed by the processor, enables the functions / operations specified in the flow chart and / or the block diagram to be implemented. The computer program can be executed completely on the machine, partially on the machine, as an independent software package, partially on the machine and partially on a remote machine, or completely on a remote machine or server.

[0326] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to execute a base identification method, the method comprising:

[0327] Obtaining a first mapping relationship between correct and incorrect classification information of initial base recognition results of a sequencing round and first sequencing information, wherein the first sequencing information includes first initial base recognition information based on a target sequencing round, and the first initial base recognition information includes initial base recognition information of a consecutive sequencing rounds, and the a consecutive sequencing rounds at least include the target sequencing round;

[0328] Based on the first mapping relationship and the first sequencing information of the sequencing round to be processed, correct or incorrect classification information of the initial base recognition result of the sequencing round to be processed is determined.

[0329] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0330] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0331] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0332] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0333] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0334] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A base recognition method, characterized in that: include: Obtaining a first mapping relationship between correct and incorrect classification information of initial base recognition results of a sequencing round and first sequencing information, wherein the first sequencing information includes first initial base recognition information based on a target sequencing round, and the first initial base recognition information includes initial base recognition information of a consecutive sequencing rounds, and the a consecutive sequencing rounds at least include the target sequencing round; Based on the first mapping relationship and the first sequencing information of the sequencing round to be processed, correct or incorrect classification information of the initial base recognition result of the sequencing round to be processed is determined.

2. The method according to claim 1, characterized in that The determining, based on the first mapping relationship and the first sequencing information of the sequencing round to be processed, the correct or incorrect classification information of the initial base recognition result of the sequencing round to be processed includes: Obtaining first sequencing information of the sequencing round to be processed; Based on the first sequencing information and the first mapping relationship, correct and incorrect classification information of the initial base recognition result of the sequencing round to be processed is determined.

3. The method according to claim 1 or 2, characterized in that: The first initial base identification information includes initial base identification information of a consecutive sequencing rounds, and the a consecutive sequencing rounds at least include the designated sequencing round; The continuous a sequencing rounds include the target sequencing round, and at least one of the following two situations: n1 sequencing rounds of base extension reaction prior to the target sequencing round, m1 sequencing rounds that perform base extension reactions later than the target sequencing round, Wherein, n1 and m1 are each independently an integer greater than or equal to 0, and m1 and n1 are not 0 at the same time; Optionally, the first initial base identification information includes a sequence of initial base identification information of consecutive a sequencing rounds; Optionally, the first initial base identification information also includes initial base identification information of the target sequencing round; Optionally, the first initial base identification information includes initial base identification information of each sequencing round in n1 sequencing rounds and / or m1 sequencing rounds, and initial base identification information of the target sequencing round; Optionally, the first initial base identification information includes initial base identification information of each sequencing round in the a consecutive sequencing rounds; Optionally, the first initial base identification information includes an initial base identification information sequence formed by at least one of n1 sequencing rounds and / or m1 sequencing rounds and a target sequencing round, and the initial base identification information of the target sequencing round; Optionally, the first initial base recognition information is the photoelectric signal intensity generated during the base extension reaction; Optionally, the photoelectric signal intensity includes optical signal intensities of multiple optical signal channels, wherein the multiple optical signals respectively correspond to multiple base types participating in the base extension reaction; Optionally, the optical signal intensity is the optical signal intensity with the largest intensity among multiple optical signal channels; Optionally, the optical signal intensity is a fluorescence signal intensity; Optionally, the fluorescence signal intensity includes any one of an initial fluorescence signal intensity and a corrected fluorescence signal intensity; Optionally, the correction includes at least one of background correction, crosstalk correction, reaction asynchrony correction and normalization processing.

4. The method according to any one of claims 1 to 3, characterized in that: The first mapping relationship is obtained by training a pre-constructed base recognition result correct and incorrect classification model; Optionally, the training of the base recognition result correct and incorrect classification model includes: Acquire a standard sample and a sequence of the standard sample; Performing a base extension reaction on the standard sample to determine second sequencing information of a target sequencing round and an initial base recognition result of base extension of the target sequencing round, wherein the second sequencing information at least includes the first sequencing information; Determining the correct or incorrect category of the initial base recognition of the target sequencing round based on the initial base recognition result of the base extension of the target sequencing round and the standard base; Using the second sequencing information as a feature value and the correct or incorrect category of the initial base recognition sequence as a target value, constructing a training sample; Based on the training samples, a pre-constructed base recognition result correct / error classification model is trained to obtain a trained base recognition result correct / error classification model; Optionally, the true or false category may be a binary category or a multi-category category with a number of categories greater than 2; Optionally, the correct and incorrect categories include the following two categories: a correct category of the initial base recognition result and an incorrect category of the initial base recognition result; Optionally, the correct and incorrect categories include the following five categories: The correct category of the initial base calling result; A base error category; T or U base error category; G base error category; C base error category.

5. The method according to any one of claims 1 to 4, characterized in that: The base recognition method further comprises: Determining the base recognition result of the sequencing round to be processed based on the correct and incorrect classification information of the initial base recognition result; Optionally, the determining the base recognition result of the sequencing round to be processed based on the correct or incorrect classification information of the base recognition includes: Acquire a second mapping relationship between the standard base of the sequencing round and the third sequencing information, wherein the third sequencing information includes the second initial base recognition information based on the target sequencing round and the correctness or incorrectness classification information of the initial base recognition result; Determine a base recognition result of the sequencing round to be processed based on the second mapping relationship and the third sequencing information of the sequencing round to be processed; Optionally, determining the initial base recognition result of the sequencing round to be processed based on the second mapping relationship and the third sequencing information of the sequencing round to be processed includes: Obtaining third sequencing information of the sequencing round to be processed; Based on the third sequencing information and the second mapping relationship, determining the probabilities of the four base types of the sequencing round to be processed, and taking the base type with the greatest probability as the base recognition result of the sequencing round to be processed; Optionally, the second initial base identification information includes initial base identification information of the target sequencing round; The second initial base recognition information is the photoelectric signal intensity generated when the target sequencing round performs a base extension reaction; Optionally, the photoelectric signal intensity includes optical signal intensities of multiple optical signal channels, wherein the multiple optical signals respectively correspond to multiple base types participating in the base extension reaction; Optionally, the optical signal intensity is the optical signal intensity with the largest intensity among multiple optical signal channels; Optionally, the optical signal intensity is a fluorescence signal intensity; Optionally, the fluorescence signal intensity includes any one of an initial fluorescence signal intensity and a corrected fluorescence signal intensity; Optionally, the correction includes at least one of background correction, crosstalk correction, reaction asynchrony correction and normalization processing.

6. The method according to claim 5, characterized in that The third sequencing information also includes at least one of the following information: The E value from the target sequencing round; The sequence determined by the E value of each sequencing round in b consecutive sequencing rounds; The photoelectric signal intensity generated by each sequencing round in b consecutive sequencing rounds, Wherein, the E value is the ratio of the maximum value of the photoelectric signal intensity of the four base types in the optical detection channel to the sum of the photoelectric signal intensity of the four base types; Optionally, the b consecutive sequencing rounds are selected from at least one of the following: Including a target sequencing round and n2 sequencing rounds that perform base extension reactions prior to the target sequencing round; or Including a target sequencing round and m2 sequencing rounds that perform a base extension reaction later than the target sequencing round; or Including a target sequencing round, n2 sequencing rounds that complete sequencing before the target sequencing round, and m2 sequencing rounds that complete sequencing after the target sequencing round; Wherein, n2 and m2 are each independently an integer greater than or equal to 0, and m2 and n2 are not 0 at the same time.

7. The method according to any one of claims 5-6, characterized in that: The base recognition method further comprises: Based on the base recognition result, determining the base recognition error rate of the sequencing round to be processed; Optionally, determining the base recognition error rate of the sequencing round to be processed based on the base recognition result includes: Acquire a third mapping relationship between fourth sequencing information of a sequencing round and the base recognition error rate, wherein the fourth sequencing information includes recognition probability scores of four base types; Determining a base recognition error rate of the sequencing round to be processed based on the third mapping relationship and fourth sequencing information of the sequencing round to be processed; Optionally, the fourth sequencing information further includes correctness or error classification information of the initial base recognition result, and the correctness or error classification information includes the following five categories: The correct category of the initial base calling result; A base error category; T or U base error category; G base error category; C base error category.

8. A base recognition device, characterized in that: include: A first mapping relationship acquisition module is configured to acquire a first mapping relationship between the correct and incorrect classification information of the initial base recognition result of the sequencing round and the first sequencing information, wherein the first sequencing information includes the first initial base recognition information based on the target sequencing round, and the first initial base recognition information includes the initial base recognition information of a consecutive sequencing rounds, and the a consecutive sequencing rounds at least include the target sequencing round; The right-or-wrong classification information determination module is used to determine the right-or-wrong classification information of the initial base recognition result of the sequencing round to be processed based on the first mapping relationship and the first sequencing information of the sequencing round to be processed.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the base recognition method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the base recognition method according to any one of claims 1 to 7 when executed.

Citation Information

Cited By

  • System and method for automatically generating machine learning model

    CN120220826A