Low-frequency mononucleotide variation detection method and device

Abstract

The invention provides a low-frequency mononucleotide variation detection method and device, and relates to the technical field of nucleic acid sequencing, and the method comprises the following steps: obtaining sequencing data and a corresponding reference genome, comparing the sequencing data to the reference genome, and carrying out stacking analysis processing to obtain comparison information; potential variation positions are extracted based on the comparison information to serve as candidate sites; extracting sequence information covering the candidate sites as candidate sequences; removing the same gap vacancies according to the coordinate positions; the alignment treatment of the basic groups in different directions is repeatedly carried out until the alignment treatment results of two adjacent times do not change, so that an optimized comparison result is obtained; and outputting a low-frequency variation detection result. According to the method, false positive results are reduced through multi-sequence alignment and re-alignment, the result accuracy is ensured, the minimum number of reads supported by mutation is not limited, the identification sensitivity of low-frequency mutation is kept, and missing detection is avoided; low-frequency variation sites are comprehensively recognized through accumulation analysis, and the method is suitable for third-generation sequencing data.

Description

Technical Field

[0001] The present invention relates to the technical field of nucleic acid sequencing, and more specifically, to a method and device for detecting low-frequency single nucleotide variation. Background Art

[0002] With the development of genomics, single nucleotide variation (SNV) detection technology has played a key role in biomedical research. SNV is the most common type of variation in the genome and is of great significance to the research and diagnosis of many diseases. The development of genetic testing technology has gone through the process from traditional Sanger sequencing to high-throughput second-generation sequencing (NGS), and then to the latest third-generation sequencing (TGS). Among them, the third-generation sequencing technology represented by PacBio's HiFi sequencing and NanoPore's Q20 sequencing technology has shown great potential in genome variation detection due to its long read length and high accuracy.

[0003] With its high throughput and low cost, the second-generation sequencing technology has become a mainstream tool for biomedical research. Current variant detection mainly relies on the second-generation sequencing technology. Data analysis software based on the second-generation sequencing, such as DeepVariant and VarDict, has been widely used in genomic variant detection. These software mainly extract potential variant sites by aligning sequencing data to the reference genome, and perform detailed analysis and filtering to predict and identify mutations. However, the low-frequency variant detection method based on the second-generation sequencing data requires that the input data has extremely high accuracy, but the second-generation sequencing technology itself is prone to high error rates when reading long fragments, so it performs poorly in the detection of low-frequency variants, especially somatic low-frequency variants. For example, when DeepVariant extracts candidate variant sites and predicts mutations using a convolutional neural network (CNN) model, some low-depth variants may be missed due to the filtering step. When processing the second-generation data, VarDict still cannot completely eliminate false positive results caused by sequencing errors despite multiple sequence alignment and re-alignment of variant sites. This leads to sensitivity and accuracy problems in low-frequency variant detection.

[0004] Specifically, DeepVariant performs certain filtering when extracting candidate variant sites to avoid too many candidate sites affecting the reliability of calculations, but it may also miss some low-frequency variants. When predicting mutations, this method further filters the results to reduce the proportion of false positives, but this also means that some real low-frequency variants may be ignored. Similarly, VarDict only applies to second-generation data and lacks effective response measures for error handling of third-generation data. This method handles sequencing errors through multiple sequence alignment and re-alignment, but it is still difficult to reduce false positive results while maintaining high sensitivity.

[0005] In summary, existing low-frequency variant detection technologies face challenges when processing third-generation sequencing data. These technologies are mainly designed for second-generation sequencing data and cannot fully utilize the advantages of third-generation sequencing technology, especially in terms of long read length and high accuracy. Existing filtering and alignment methods show low sensitivity and accuracy in detecting low-frequency variants, and it is difficult to meet the current needs for low-frequency variant detection in fields such as tumors and mitochondrial diseases. Therefore, it is necessary to develop new detection methods to better utilize third-generation sequencing technology and improve the performance and reliability of low-frequency variant detection. Summary of the invention

[0006] In view of the above technical problems, the present invention provides a method for detecting low-frequency single nucleotide variations, comprising:

[0007] S100, acquiring sequencing data and a corresponding reference genome, and aligning the sequencing data to the reference genome to obtain alignment information;

[0008] S200, extracting potential variation positions as candidate sites based on the comparison information;

[0009] S300, extracting sequence information covering the candidate site as a candidate sequence;

[0010] S400, performing position alignment on the same parts of bases between the candidate sequences to obtain a multiple sequence alignment result; and removing the same gaps from the sequences in the multiple sequence alignment result according to the coordinate positions, and repeatedly performing the alignment process of bases in different directions until the results of two adjacent alignment processes are unchanged, thereby obtaining an optimized alignment result;

[0011] S500: Outputting a low-frequency variation detection result based on the optimized comparison result.

[0012] Preferably, the S100 includes:

[0013] S110, obtaining the sequencing data and the reference genome, performing comparison analysis on the sequencing data based on the reference genome, and obtaining a comparison result file containing the location of the sequencing data in the reference genome; wherein the comparison result file is a file after sorting and indexing;

[0014] S120, performing stacking and statistical processing on the comparison result file to obtain the comparison information.

[0015] Preferably, the S120 includes:

[0016] S121, dividing the reference genome into multiple analysis windows;

[0017] S122, performing cigar character string analysis on each base of the sequencing data on the analysis window to obtain an analysis result; wherein the alignment event type in the analysis result includes at least one of a homomatch, a mismatch, an insertion, and a deletion;

[0018] S123, using the analysis result as the comparison information based on the base position corresponding to each base.

[0019] Preferably, the S200 includes:

[0020] S210, counting the number of the alignment event types corresponding to each base position;

[0021] S220, determining whether the number of the alignment event types is not less than 2, and the number of reads supported by each alignment event type is not less than the base position of the preset filtering standard;

[0022] S230, if yes, the base position is a potential variation position, and the base position is used as the candidate site;

[0023] S240, if not, the base position is not a potential variation position and is not used as the candidate site;

[0024] In S220, the method for determining the preset filtering standard includes:

[0025] S221, obtaining the sequencing depth of the current region;

[0026] S222, if the sequencing depth is not less than 10, the preset filtering criteria is 2;

[0027] S223, if the sequencing depth is less than 10, the preset filtering standard is 1;

[0028] After S200, the method further includes:

[0029] S600, determining the position interval between two adjacent candidate sites;

[0030] S700: Merge two adjacent candidate sites whose position interval is less than 2 into one candidate site.

[0031] Preferably, the S300 includes:

[0032] S310, obtaining the coordinate position of the candidate site;

[0033] S320, taking the coordinate position as the center, extracting the base corresponding to the coordinate position and the adjacent bases with a preset number of interval positions before and after to form a sequence covering the candidate site, and using the sequence as the candidate sequence.

[0034] Preferably, the S400 includes:

[0035] S410, comparing the candidate sequences in pairs to determine the identical partial bases in the sequences as identical components;

[0036] S420, using the same components as a reference, using gaps to fill in unmatched positions of the candidate sequences, so that the candidate sequences are aligned through the same components to obtain the multiple sequence alignment result; wherein the total length of the bases of each sequence in the multiple sequence alignment result is equal to the total length of the gaps;

[0037] S430, deleting gaps at the same position between sequences in the multiple sequence alignment result;

[0038] S440, performing alignment processing of bases in different directions to obtain a first optimization result;

[0039] S450, performing the alignment process of the bases in different directions on the first optimization result again to obtain a second optimization result;

[0040] S460, determining whether the second optimization result is consistent with the first optimization result;

[0041] S470: If yes, use the first optimization result or the second optimization result as the optimization comparison result;

[0042] S480: If not, the second optimization result is returned to execute S450.

[0043] Preferably, the alignment process of the bases in different directions in S450 includes:

[0044] S451, performing left alignment processing and right alignment processing on each candidate sequence to obtain an arrangement result;

[0045] S452, deleting gaps at the same position between sequences in the sorting result;

[0046] S453, repairing the gaps at the ends of each candidate sequence.

[0047] Preferably, the left alignment process includes:

[0048] In the multiple sequence alignment result, a gap in a candidate sequence of the multiple sequence alignment result is determined; and the candidate sequence is used as a first original sequence, and the base adjacent to the right of the gap in the first original sequence is used as a right base; another candidate sequence in the multiple sequence alignment result to be compared is used as a first comparison sequence, and the base corresponding to the position of the gap in the first comparison sequence is used as a first comparison base;

[0049] If the right base is the same as the first comparison base, the position of the right base in the first original sequence is swapped with the position of the gap so that the position of the right base in the first original sequence is the same as the first comparison base in the first comparison sequence and the base is the same;

[0050] The right alignment process includes:

[0051] In the multiple sequence alignment result, a gap in a candidate sequence of the multiple sequence alignment result is determined; and the candidate sequence is used as a second original sequence, and the base adjacent to the left of the gap in the second original sequence is used as a left base; another candidate sequence in the multiple sequence alignment result to be compared is used as a second comparison sequence, and the base corresponding to the position of the gap in the second comparison sequence is used as a second comparison base;

[0052] If the left base is the same as the second comparison base, the position of the left base in the second original sequence is swapped with the gap so that the position of the left base in the first original sequence is the same as the second comparison base in the second comparison sequence and the base is the same.

[0053] Preferably, the S500 includes:

[0054] S510, taking the mutated bases in the sequence of the optimized alignment result as mutation information, and merging and removing redundancy of the mutation information;

[0055] S520, filtering the mutation information after the combined de-redundancy processing according to a preset filtering standard to obtain the low-frequency mutation detection result.

[0056] In addition, the present invention also provides a low-frequency single nucleotide variation detection device, comprising:

[0057] A data comparison module, used to obtain sequencing data and a corresponding reference genome, and compare the sequencing data to the reference genome to obtain comparison information;

[0058] A site extraction module, used for extracting potential variation positions as candidate sites based on the comparison information;

[0059] A site coverage module, used to extract sequence information covering the candidate site as a candidate sequence;

[0060] A multiple alignment module is used to align the same parts of the bases between the candidate sequences to obtain a multiple sequence alignment result; and remove the same gaps in the sequences in the multiple sequence alignment result according to the coordinate positions, and repeatedly align the bases in different directions until the results of two adjacent alignment processes are unchanged, thereby obtaining an optimized alignment result;

[0061] A result output module is used to output a low-frequency variation detection result based on the optimized comparison result.

[0062] The present invention provides a method and device for detecting low-frequency single nucleotide variations. The method significantly improves the accuracy and sensitivity of detection by improving the alignment and analysis process. Through multiple sequence alignment and re-alignment, it is possible to effectively distinguish between true variations and pseudo-variations, reduce false positive results caused by sequencing errors, and thus improve the accuracy of the results; there is no minimum requirement for the number of reads that support mutations, thereby maintaining the sensitivity of identifying low-frequency mutations and not missing mutations; at the same time, stacking analysis is used to comprehensively identify low-frequency variation sites, which is particularly suitable for the processing of third-generation sequencing data; and the systematic and automated process improves the detection efficiency, providing more reliable and efficient technical support for genomic research and clinical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is a schematic diagram of the structure of the hardware operating environment of the low-frequency single nucleotide variation detection system of the present invention;

[0064] Figure 2 A schematic diagram of a process for detecting a low-frequency single nucleotide variation according to an embodiment of the present invention;

[0065] Figure 3 Schematic diagram of a preferred implementation of step S100 in the method for detecting low-frequency single nucleotide variation of the present invention;

[0066] Figure 4 Schematic diagram of a preferred implementation of step S120 in the method for detecting low-frequency single nucleotide variation of the present invention;

[0067] Figure 5 Schematic diagram of a preferred implementation of step S200 in the method for detecting low-frequency single nucleotide variation of the present invention;

[0068] Figure 6 A schematic flow chart of a preferred implementation of determining the preset filtering criteria in step S220 of the method for detecting low-frequency single nucleotide variations of the present invention;

[0069] Figure 7 Schematic diagram of the overall process of the low-frequency single nucleotide variation detection method of the present invention including steps S600 and S700;

[0070] Figure 8 Schematic diagram of a preferred implementation of step S300 in the method for detecting low-frequency single nucleotide variation of the present invention;

[0071] Fig. 9 Schematic diagram of a preferred implementation of step S400 in the method for detecting low-frequency single nucleotide variation of the present invention;

[0072] Fig.10 Schematic diagram of a preferred implementation of step S420 in the method for detecting low-frequency single nucleotide variation of the present invention;

[0073] Fig.11 Schematic diagram of a preferred implementation of step S500 in the method for detecting low-frequency single nucleotide variation of the present invention;

[0074] Fig.12 Schematic diagram of module connection of the low-frequency single nucleotide variation detection device of the present invention. DETAILED DESCRIPTION

[0075] The embodiments of the present invention are described in detail below, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions.

[0076] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly defined.

[0077] In the present invention, unless otherwise clearly defined, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. Those skilled in the art can understand the specific meanings of the above terms in the present invention according to the specific circumstances.

[0078] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0079] like Figure 1 , which is a schematic diagram of the structure of the hardware operating environment of the terminal involved in the embodiment of the present invention.

[0080] The low-frequency single nucleotide variation detection system provided in the embodiment of the present invention may be a PC, or a mobile terminal device such as a smart phone, a tablet computer or a portable computer. The low-frequency single nucleotide variation detection system may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005 and a communication bus 1002. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen, an input unit such as a keyboard, and a remote control. The optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory, or a stable memory, such as a disk memory. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0081] Those skilled in the art will understand that Figure 1 The low-frequency single nucleotide variation detection system shown in the figure does not constitute a limitation thereto, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently. Figure 1 As shown, the memory 1005 as a computer-readable storage medium may include an operating system, a data interface control program, a network connection program, and a low-frequency single nucleotide variation detection program.

[0082] Reference Figure 2 In an embodiment of the present invention, the low-frequency single nucleotide variation detection method comprises:

[0083] Step S100, obtaining sequencing data and a corresponding reference genome, and aligning the sequencing data to the reference genome to obtain alignment information.

[0084] Step S200: extracting potential variation positions as candidate sites based on the comparison information.

[0085] Step S300, extracting sequence information covering the candidate site as a candidate sequence.

[0086] Step S400, aligning the same parts of the bases between the candidate sequences to obtain a multiple sequence alignment result; and removing the same gaps in the sequences in the multiple sequence alignment result according to the coordinate positions, and repeating the alignment process of the bases in different directions until the results of two adjacent alignment processes do not change, thereby obtaining an optimized alignment result.

[0087] Step S500: outputting a low-frequency variation detection result based on the optimized comparison result.

[0088] The above step S100 is used to obtain DNA or RNA sequencing data from the sample and compare the data with the reference genome. Through the comparison, possible mutation locations can be found.

[0089] The sequencing data may be sequencing data obtained using third-generation sequencing technology, such as sequencing data obtained based on Pacbio sequencing.

[0090] "Reference genome" refers to a standardized genome sequence used as a comparison object in genomic research and analysis. It is usually composed of the genome sequences of one or several individuals in a species, and is used as a benchmark to study the genomic variations of other individuals. For example, the reference genome can be hg19. hg19 is a version of the human genome reference sequence, representing the Human Genome Reference Consortium (HGRC). Specifically, a sequencer can be used to generate sequencing data (such as FASTQ format or other formats), and then alignment software (such as BWA, Bowtie) can be used to align sequencing reads to the reference genome. This alignment can better understand the coverage and depth information in the alignment results, thereby generating an alignment file (such as BAM format or other formats), which contains the location of the sequencing reads on the genome and alignment quality information. For example, the sequencing data of a sample is aligned to the human reference genome hg19 using BWA software to generate an alignment file sample.bam.

[0091] The above step S200 is used to identify the position where the variation may exist from the comparison information. Through the preliminary screening of this step, the scope of detailed analysis can be narrowed down, and the efficiency of subsequent analysis can be improved. Specifically, the base positions different from the reference genome can be found by analyzing the comparison file, and these positions are recorded as candidate variation sites, thereby generating a candidate site list, including the position where the variation may exist and related information (such as comparison depth, variation frequency, etc.). Candidate sites can be extracted using variation detection software such as GATK or Varscan. For example, the HaplotypeCaller tool of GATK is used to extract potential variation sites from the comparison file sample.bam, and generate a candidate site file sample.vcf.

[0092] The above step S300 is used to extract the read information covering the candidate variant site from the sequencing data. Specifically, all reads covering each candidate site can be extracted from the alignment file, and the sequence information of these reads can be recorded to generate a candidate sequence data set containing all read sequences covering the candidate site. By extracting these sequences, the variation of each candidate site can be analyzed in more detail.

[0093] For example, the view command of Samtools can be used to extract the reads covering the candidate site. Samtools is used to extract all reads covering a candidate site chr1:1000000 from the alignment file sample.bam to generate a candidate sequence file reads_at_chr1_1000000.bam.

[0094] In the above step S400, firstly, the candidate sequences are subjected to the removal of identical gaps according to the coordinate positions to obtain a multiple sequence alignment result;

[0095] Then, based on the multiple sequence alignment result, the alignment process of bases in different directions is repeated until the results of two adjacent alignment processes are unchanged, thereby obtaining an optimized alignment result.

[0096] This step is used to perform a more rigorous alignment and re-alignment on the extracted candidate sequences to optimize the alignment results.

[0097] By removing the same gaps and repeating the base alignment process in different directions, that is, multi-sequence base alignment optimization processing, the location and type of the variation can be determined more accurately and false positive results can be reduced.

[0098] Specifically, for example, a multiple sequence alignment tool (such as ClustalW, MUSCLE) can be used to align the candidate sequences, and then a re-alignment tool can be used to perform a detailed re-alignment to ensure the accuracy of the alignment result, thereby generating an optimized alignment result containing accurate alignment information of the candidate sequences.

[0099] The above process of removing the same gaps based on the coordinate position is because a variation is the difference in base positions after alignment of two or more sequences. The variation here refers to the base at that position, while the gaps between different sequences are not included in the "variation". The gaps have no meaning for subsequent processing and operations, so they are deleted.

[0100] In the preferred embodiment of the method provided in the present application, the process of removing the same gaps according to the coordinate positions is performed multiple times because if it is only performed once, there is a possibility that some gaps that should be deleted may be missed, in order to improve the accuracy of data processing.

[0101] The above step S500 is used to generate the final low-frequency variation detection result based on the optimized comparison result. Specifically, the optimized comparison file can be analyzed to identify and record all low-frequency variation sites and their detailed information (such as variation type, variation frequency, quality score, etc.), thereby generating a variation detection report containing all identified low-frequency variations and related information.

[0102] Reference Figure 3 In a preferred embodiment of the present invention, step S100 includes:

[0103] Step S110, obtaining the sequencing data and the reference genome, performing comparison analysis on the sequencing data based on the reference genome, and obtaining a comparison result file containing the location of the sequencing data in the reference genome; wherein the comparison result file is a file after sorting and indexing.

[0104] Step S120, performing stacking and statistical processing on the comparison result file to obtain the comparison information.

[0105] The comparison analysis in the above step S110 can be performed by aligning the third-generation HiFi data based on Pacbio sequencing to the reference genome (hg19) using minimap2. The sorting process can be performed by samtools. The sorted comparison result file can ensure that all sequencing reads are arranged in the order of the reference genome, which is convenient for subsequent analysis and processing. The sorted file makes it more efficient when performing region-specific queries. For example, when performing variation detection or coverage analysis, it is often necessary to access specific regions of the genome, and the sorted files can significantly speed up this process.

[0106] It should be noted that in operations such as variant detection and accumulation statistics processing, many tools require the input files to be sorted. Sorting ensures that these tools can process data quickly and effectively. When processing sorted files, memory can be used more efficiently, avoiding memory waste and processing bottlenecks caused by random access to large amounts of unsorted data.

[0107] The index construction can be performed by any one or more of open source software such as minimap2, samtools, etc. The indexed file provides a fast random access method for the comparison file, so that the required specific genome position can be quickly located in a large file. When performing parallel computing, the index allows multiple threads to access different genome regions simultaneously without having to rely on sequential scanning of the entire file. The indexed file makes it faster when performing queries in specific regions, because the indexed file records the offset of each region, so that the tool can jump directly to the required region for processing. Many genomics analysis tools, such as Samtools, GATK, etc., rely on index files to improve processing speed and efficiency. Index files enable these tools to process large-scale genomic data faster.

[0108] The above-mentioned pileup statistical processing, namely "pileup statistics", is based on pileup analysis (Pileup Analysis) or pileup processing method, which refers to an operation in genome sequencing data analysis, especially when processing aligned data. Pileup itself usually refers to stacking all sequencing reads (reads) aligned to a specific position in the genome so that the base conditions of these reads at that position can be observed and analyzed. This operation is often used for variation detection, including the identification of single nucleotide polymorphisms (SNVs) and insertions and deletions (Indels). Specifically, it processes the sequencing data aligned to the reference genome to count the sequencing coverage, variation information, and sequencing quality information at each genomic position.

[0109] For further reference, Figure 4 The step S120, performing stacking statistics processing on the comparison result file to obtain the comparison information, includes:

[0110] S121, dividing the reference genome into multiple analysis windows;

[0111] S122, performing cigar character string analysis on each base of the sequencing data on the analysis window to obtain an analysis result; wherein the alignment event type in the analysis result includes at least one of a homomatch, a mismatch, an insertion, and a deletion;

[0112] S123, using the analysis result as the comparison information based on the base position corresponding to each base.

[0113] In the above step S121, the reference genome sequence can be segmented according to a fixed length or specific conditions to obtain multiple analysis windows, each window containing a genome sequence of a certain length. The segmentation process helps to parallelize the analysis and improve the processing efficiency. Specifically, a script can be written or existing tools can be used to segment the reference genome.

[0114] Adjacent analysis windows may be connected by sequential cutting positions, or may have overlapping regions after cutting. For example, based on the coordinate positions of the bases, the following may be true:

[0115] Table 1. Segmentation based on coordinate positions in multiple analysis windows

[0116] window Analysis Window 1 Analysis Window 2 Analysis Window 3 Analysis window n Base coordinate position 1-50 45-95 90-140 ……

[0117] In genomic analysis, when the genome is divided into multiple windows for parallel computing processing, overlapping regions are designed to ensure the integrity and accuracy of data processing.

[0118] The cigar string described in the above step S122 is a string that describes the alignment of all bases of a read after alignment (such as mismatch, insertion, etc.). The pileup statistics are pileup statistics. The parsing of the cigar string refers to analyzing the alignment event type one by one. These event types include match, mismatch, insertion and deletion. Specifically, the cigar string can be parsed by writing a script or using existing bioinformatics tools. For example, the cigar string can be parsed character by character with a Python script to parse the alignment information therein. For example, the numbers and letters in the cigar string are parsed to determine the type and position of each alignment event. The parsing results are recorded, and the specific alignment events of each base are indicated, so as to obtain the parsing results, which record the position of each base and its corresponding alignment event type (match, mismatch, insertion, deletion).

[0119] This step can accurately identify various types of variants, improve the accuracy of variant detection, and facilitate precise statistics and analysis in subsequent steps.

[0120] The significance of setting overlapping areas for multiple analysis windows in this step is:

[0121] First, during the genome segmentation process, there may be variations or key alignment information at the boundaries of each window. If the overlapping regions are not considered, the information at these boundaries may be ignored or misjudged, resulting in inaccurate test results. The overlapping regions can ensure that each window can fully contain the information near the boundaries during analysis, avoiding the influence of boundary effects.

[0122] Second, the variation in the genome may span multiple windows, especially insertion and deletion (INDEL) variation. If there is no overlap between windows, the variation across windows may not be correctly identified and processed. The overlap ensures that the variation across windows is fully captured and analyzed in at least one window.

[0123] Third, when performing variant detection and stacking analysis, overlapping regions can provide redundant information to improve the reliability and accuracy of detection. Even if some reads within a window cause inaccurate analysis results due to noise or other reasons, the additional data provided by the overlapping regions can help verify and correct these results.

[0124] In short, this overlapping area setting can ensure that the analysis data in each window is complete and covers all key areas without omission, especially the important information at the window boundary. It improves the accuracy and continuity of variant detection and avoids variant omission or misjudgment caused by window segmentation. In parallel computing, overlapping areas help coordinate data consistency between different threads or computing units and reduce processing conflicts at the boundary.

[0125] In step S123, the analysis result is based on the base position corresponding to each base as the comparison information. Specifically, the analysis result can be written into a dictionary based on the base position corresponding to each base, and a dictionary is created for the purpose, wherein the key is the base position and the value is the comparison event type of the position. The analysis results are traversed, and each comparison event is written into the dictionary, thereby generating a dictionary to record each base position and its corresponding comparison event type. The dictionary structure facilitates rapid query and statistics of the comparison events of each base position, improves the efficiency of data processing, and contributes to the comprehensive analysis and statistics of the comparison information in subsequent steps.

[0126] Reference Figure 5 In a preferred embodiment of the present invention, the step S200 preferably includes:

[0127] S210, counting the number of the alignment event types corresponding to each base position;

[0128] S220, determining whether the number of the alignment event types is not less than 2, and the number of reads supported by each alignment event type is not less than the base position of the preset filtering standard;

[0129] S230, if yes, the base position is a potential variation position, and the base position is used as the candidate site;

[0130] S240, if not, the base position is not a potential variation position and is not used as the candidate site;

[0131] refer to Figure 6 In step S220, the method for determining the preset filtering standard preferably includes:

[0132] S221, obtaining the sequencing depth of the current region;

[0133] S222, if the sequencing depth is not less than 10, the preset filtering criteria is 2;

[0134] S223, if the sequencing depth is less than 10, the preset filtering standard is 1;

[0135] The above step S210 can record the comparison of each base position in detail, which is helpful for identifying the variant position and can ensure that only the variant position with sufficient supporting evidence is considered as a potential variant, thereby improving the accuracy and reliability of the detection.

[0136] The sequencing depth mentioned above refers to the number of times a specific position is covered by sequencing in genome sequencing. It is an important indicator of the quality and reliability of sequencing data. A high sequencing depth means that the position is covered by more reads, thereby increasing the accuracy and reliability of the sequence information at that position. In the above steps, the sequencing depth is used to determine the filtering criteria for variation detection. Specifically, the sequencing depth determines the minimum requirement for the number of reads that support variation when determining whether a base position is a potential variation site. If the sequencing depth of a region is greater than or equal to 10 reads, the filtering criteria for variation detection is set to the number of reads that support variation to be no less than 2. If the sequencing depth of a region is less than 10 reads, the filtering criteria for variation detection is set to the number of reads that support variation to be no less than 1.

[0137] refer to Figure 7 , after step S200, the following steps are also included:

[0138] Step S600, determining the position interval between two adjacent candidate sites;

[0139] Step S700: merge two adjacent candidate sites whose position interval is less than 2 into one candidate site.

[0140] The above step S600 can generate candidate variation site spacing information, providing a basis for merging adjacent candidate sites. Specifically, the spacing between adjacent sites can be calculated by traversing the candidate variation site list.

[0141] The above step S700 can merge the candidate sites that meet the conditions, thereby reducing redundancy and improving the accuracy of variation detection.

[0142] refer to Figure 8 In a preferred embodiment of the present invention, step S300 comprises:

[0143] Step S310, obtaining the coordinate position of the candidate site;

[0144] Step S320, taking the coordinate position as the center, extracting the base corresponding to the coordinate position and the adjacent bases with a preset number of interval positions before and after to form a sequence covering the candidate site, and using the sequence as the candidate sequence.

[0145] The coordinate information in step S310 includes the chromosome number and the genome position. This step lays the foundation for extracting relevant sequence information in the subsequent step. Specifically, the candidate site list can be traversed to read and record the coordinates of each candidate site. For example, a dictionary structure can be used to store these coordinate information.

[0146] The extraction in step S320 can be performed as follows: read the sequencing reads in the alignment file to obtain all reads covering the region, integrate the sequence information of these reads to form a candidate sequence covering the candidate site and its adjacent region, thereby generating sequence information covering the candidate site, which will be used as input for subsequent multiple sequence alignment.

[0147] The preset number of interval positions is the number of base positions extending in both the forward and backward directions with the coordinates of the candidate site as the center. For example, the preset number of interval positions can be 3, which means that based on the center, 3 bases are extended in both the forward and backward directions.

[0148] Take the following sequence as an example:

[0149] Table 2. Sequence information

[0150] Read sequence Candidate site corresponding base Candidate sequence 1 TGTCTTTTCATTT T TCTTT 2 TGTCTTTTCA-TT T TCTTT 3 TGTCTTTTCA-TT T TCTTT 4 TGTCTTTTCA-TT T TCTTT 5 TGTC-TTTCATTT - TC-TT 6 TGTC-TTTCATTT - TC-TT 7 TCATTTTT none Not crossed

[0151] In the above table, the coordinate position of the candidate site is 5, and the base at this position is indicated in italics; the preset number of interval positions = 2.

[0152] (1) For site 5, the sequences that “span” the candidate sites are 1, 2, 3, 4, 5, and 6, centered on the coordinate position;

[0153] (2) The obtained sequences are shown in Table 2. Among them, "-" can be retained in the candidate sequence, i.e., gap.

[0154] Reference Fig. 9 In a preferred embodiment of the present invention, the step S400 preferably includes:

[0155] S410, comparing the candidate sequences in pairs to determine the identical partial bases in the sequences as identical components;

[0156] S420, using the same components as a reference, using gaps to fill in unmatched positions of the candidate sequences, so that the candidate sequences are aligned through the same components to obtain the multiple sequence alignment result; wherein the total length of the bases of each sequence in the multiple sequence alignment result is equal to the total length of the gaps;

[0157] S430, deleting gaps at the same position between sequences in the multiple sequence alignment result;

[0158] S440, performing alignment processing of bases in different directions to obtain a first optimization result;

[0159] S450, performing the alignment process of the bases in different directions on the first optimization result again to obtain a second optimization result;

[0160] S460, determining whether the second optimization result is consistent with the first optimization result;

[0161] S470: If yes, use the first optimization result or the second optimization result as the optimization comparison result;

[0162] S480: If not, the second optimization result is returned to execute S450.

[0163] In step S420, based on the identical components determined in step S410, the identical components are aligned and a certain number of gaps are used to fill the gaps, so that the identical components in the two compared candidate sequences can correspond in position, and the total lengths of the two sequences after alignment and filling (including bases and gaps) are equal.

[0164] The multiple sequence alignment algorithm (MSA) described in the above step 410 is to arrange three or more biological sequences (usually protein, DNA or RNA sequences) together and compare them in pairs so as to identify the similarities and differences between them. Multiple sequence alignment is a commonly used technology in bioinformatics, which is used to study the evolutionary relationship, functional domain and conserved region between sequences, etc., and its purpose is to find out the conserved regions in multiple sequences, which are often related to important biological functions. By comparing the sequences, the evolutionary relationship between the sequences is inferred and a phylogenetic tree is constructed. According to the conserved region, the function of the unknown sequence can be predicted.

[0165] The POA multiple sequence alignment algorithm can be used. POA (Partial Order Alignment) is a multiple sequence alignment algorithm based on partially ordered graphs. Unlike traditional linear sequence alignment, POA can efficiently handle multiple sequence alignment problems containing insertions and deletions (INDELs).

[0166] In step S410, all candidate sequences are aligned in pairs using a multiple sequence alignment algorithm, and identical partial bases are determined as identical components.

[0167] For example, sequence 1 is: ATTGAGAGGAT; sequence 2 is TTGAGAGATT; the two sequences are compared, where, based on the same direction, such as starting from the left side, the same partial bases, that is, the same components, are: (1) TTGAGA, (2) AG, and (3) T.

[0168] For example, after the aforementioned sequence 1 and sequence 2 are processed in step S420, the multiple sequence alignment result obtained is:

[0169] Table 3. Candidate sequences before and after alignment

[0170] Read Before alignment After alignment 1 ATTGAGAGGAT ATTGAG-AG--GAT- 2 TTGAGAAGATT -TTGAGAAGA---TT

[0171] In the above table, the candidate sequences are aligned according to the coordinate positions so that the same components such as (1) TTGAGA, (2) AG, and (3) T correspond to each other, and gaps are used to fill the positions to achieve the alignment effect.

[0172] In step S430, after obtaining the multiple sequence alignment result, it is necessary to delete the gaps between the sequences in the multiple sequence alignment result.

[0173] Each sequence may contain a different number of gaps, i.e., gaps, "-". Specifically, in the multiple sequence alignment result, one of the sequences may be used as a reference sequence, and then all other sequences may be compared with each base in the reference sequence, and a one-to-one comparison of the same positions is performed based on the positions of the bases and gaps in each candidate sequence. If there are gaps in the same positions in other candidate sequences and the reference sequence, they are deleted simultaneously.

[0174] For example, in the multiple sequence alignment structure, there are two sequences to be compared:

[0175] Read1 (as reference sequence): ATTGAG-AG--GAT-;

[0176] Read2 (other candidate sequences): -TTGAGAAGA---TT;

[0177] When this step is used for processing, a comparison is performed. The positions of the gaps in Read1 are 7, 10, 11, and 15, respectively; and the positions of the gaps in Read2 are 1, 11, 12, and 13. The parts that are in the same position and are also gaps, that is, the "-" in position 11, are deleted synchronously. After processing, the result is:

[0178] Read1 (reference sequence): ATTGAG-AG-GAT-;

[0179] Read2 (other candidate sequence): -TTGAGAAGA--TT.

[0180] The first candidate sequence in the multiple sequence alignment result can be selected as the reference sequence, and all other candidate sequences will be compared one-to-one with each base in the reference sequence. During the comparison process, if there are gaps in both sequences at the same position, these gaps are deleted. By deleting common gaps, the comparison between sequences is more focused on the actual existing bases, reducing misunderstandings caused by gaps. This method can effectively reduce the impact of incorrect gap comparisons on the results, thereby improving the accuracy and reliability of mutation detection.

[0181] In step S440 and step S450, the alignment processing of bases in different directions may be to perform left alignment processing and right alignment processing on each sequence respectively.

[0182] In the present application, after deleting the gaps in the same position in the aforementioned step S430, a "re-alignment" is performed, that is, the "alignment of bases in different directions" is repeated to obtain the results respectively, and then step S460 is used to determine whether the results of the re-alignment are consistent. If they are consistent and there is no change, it can be determined that the optimization process has achieved the goal, and then one of the results of the re-alignment, that is, any one of the first optimization result and the second optimization result, can be used as the optimized alignment result; if the results of the re-alignment are inconsistent after judgment, it is necessary to continue the "alignment of bases in different directions" until the optimization results before and after are unchanged.

[0183] The "perform again" in step S450 may be for the result obtained in the previous step, i.e., the first optimization result, or the second optimization result returned in S480 which needs to be re-aligned, and the alignment process is performed again on it to obtain a new second optimization result.

[0184] When performing the alignment process in the re-alignment, the alignment process in the re-alignment can be performed repeatedly. The output result is taken as an optimization result, that is, the first optimization result, and the "alignment process in the re-alignment" step is performed again for the first optimization result to generate a second optimization result, and the first optimization result and the second optimization result are compared. This comparison needs to achieve a "consistent" result, that is, if the two optimization results are the same, there is no change before and after, and the data information is consistent, then the optimization is completed. Otherwise, it is necessary to return to execute step S450 until the judgment criterion is met.

[0185] refer to Fig.10 The alignment process of the bases in different directions in S450 includes:

[0186] S451, performing left alignment processing and right alignment processing on each candidate sequence to obtain an arrangement result;

[0187] S452, deleting gaps at the same position between sequences in the sorting result;

[0188] S453, repairing the gaps at the ends of each candidate sequence.

[0189] Through each base in the sequence and the gap at the corresponding position, the gap is gradually moved until all the gaps are gathered on one side. After left alignment and right alignment, two different sequence alignment results can be obtained, which helps to select the best alignment method in the subsequent steps to more accurately identify the differences between sequences. This processing method can reveal the differences and similarities in the sequences hidden due to the different gap positions, and enhance the ability to identify potential mutations in the comparison area.

[0190] In order to ensure the readability and analyzability of the results, in principle, the left alignment process can be performed first and the right alignment process can be performed later in the above two processing steps of left alignment and right alignment. In addition, on the premise that the results are readable and analyzable, the right alignment process can be performed first and the left alignment process can be performed later.

[0191] In a preferred embodiment, the left alignment process is performed first, and then the right alignment process is performed.

[0192] In step S452, after the left alignment process and the right alignment process are performed and the sorting result is obtained, the first candidate sequence in the sorting result can be selected as the reference sequence, and all other candidate sequences will be compared one-to-one with each base in the reference sequence. During the comparison process, if there are gaps in both sequences at the same position, these gaps are deleted.

[0193] By removing common gaps, the comparison between sequences is more focused on the actual bases, reducing the misunderstanding caused by gaps. This method can effectively reduce the impact of gap errors on the results, thereby improving the accuracy and reliability of mutation detection.

[0194] The above step S453 focuses on repairing the terminal gaps of each candidate sequence. Terminal gaps refer to the spaces at the beginning and end of the sequence, which may be caused by the inaccuracy of the alignment algorithm. After repairing the terminal gaps, the start and end positions of the sequence will be more accurate, which helps to correctly interpret the sequence data, especially in the boundary areas of the sequence. Repairing terminal gaps can reduce misjudgments and missed judgments, especially in variant detection in the boundary areas of the genome.

[0195] Furthermore, the left alignment process in step S451 preferably includes:

[0196] Step S4511, in the multiple sequence alignment result, determine the gap in the candidate sequence of the multiple sequence alignment result; and use the candidate sequence as the first original sequence, and the base adjacent to the right of the gap in the first original sequence as the right base; use another candidate sequence in the multiple sequence alignment result to be compared as the first comparison sequence, and use the base corresponding to the position of the gap in the first comparison sequence as the first comparison base;

[0197] Step S4512, if the right base is the same as the first comparison base, swap the position of the right base in the first original sequence with the gap so that the position of the right base in the first original sequence is the same as the first comparison base in the first comparison sequence and the base is the same.

[0198] The above-mentioned left alignment processing, when the right base of the gap gap is the same as the first comparison base, moves the position of the gap gap to the right, that is, in the first original sequence, swaps the position of the gap gap and the right base so that the right base and the first comparison base are in the same position and aligned, thereby completing the left alignment processing.

[0199] For example, the following read is left-aligned:

[0200] Table 4. Comparison before and after left alignment

[0201] Read Define Name Right base First comparison base Before treatment After treatment 1 First original sequence A / GAG-AT GAGA-T 2 First comparison sequence / A GAGATT GAGATT

[0202] (1) The gap is at position 4 in the first original sequence, so position 5 is the right base, which is A;

[0203] (2) In the first comparison sequence, position 4 corresponding to the gap, i.e., the first comparison base, is A;

[0204] (3) The right base in the first original sequence is the same as the first comparison base in the first comparison sequence, both of which are A;

[0205] (4) In the first original sequence, the right base is swapped with the gap, that is, "-A" is changed to "A-"; after the adjustment, the gap "-" in the first original sequence corresponds to the A on the right side of the first comparison sequence (the third A from the left), that is, it is adjusted to position 5; and the right base is adjusted to position 4, corresponding to the first comparison base "A" position at position 4 in the first comparison sequence, completing the left alignment.

[0206] The right alignment process in step S451 preferably includes:

[0207] Step S4513, in the multiple sequence alignment result, determine the gap in the candidate sequence of the multiple sequence alignment result; and use the candidate sequence as the second original sequence, and the base adjacent to the left of the gap in the second original sequence as the left base; use another candidate sequence in the multiple sequence alignment result to be compared as the second comparison sequence, and use the base corresponding to the position of the gap in the second comparison sequence as the second comparison base;

[0208] Step S4514, if the left base is the same as the second comparison base, swap the position of the left base in the second original sequence with the gap so that the position of the left base in the first original sequence is the same as the second comparison base in the second comparison sequence and the base is the same.

[0209] The above-mentioned right alignment process is similar to the left alignment process, that is, when the left base of the gap position GAP is the same as the second comparison base, the position of the gap position GAP is moved to the left, that is, in the second original sequence, the gap position and the left base are swapped so that the left base and the second comparison base are in the same position, thereby completing the right alignment process.

[0210] For example, the following read is right-aligned:

[0211] Table 4. Comparison before and after right alignment

[0212] Read Define Name Left base Second comparison base Before treatment After treatment 1 The second original sequence T / AATT-TA AAT-TTA 2 Second comparison sequence / T AGAGTAA AGAGTAA

[0213] (1) The gap is at position 5 in the second original sequence, so position 4 is the left base, which is T;

[0214] (2) In the second comparison sequence, position 5 corresponding to the gap, i.e., the second comparison base, is T;

[0215] (3) The left base in the second original sequence is the same as the second comparison base in the second comparison sequence, both of which are T;

[0216] (4) In the second original sequence, the left base and the gap are swapped, that is, "T-" is changed to "-T";

[0217] After the adjustment, the gap "-" in the second original sequence corresponds to the G at position 4 of the second comparison sequence, that is, it is adjusted to position 4, and the left base is adjusted to position 5, corresponding to the first comparison base T in the second comparison sequence, completing the right alignment.

[0218] Reference Fig.11 In a preferred embodiment of the present invention, the step S500 preferably includes:

[0219] Step S510, taking the mutated bases in the sequence of the optimized alignment result as mutation information, and merging and removing redundancy of the mutation information;

[0220] Step S520, filtering the mutation information after the combined de-redundancy processing according to a preset filtering standard to obtain the low-frequency mutation detection result.

[0221] The purpose of step S510 is to remove redundancy and ensure that the same mutation information at the same position is recorded only once. After this step, all redundant mutation information will be merged. For example, if two mutations are at the same position and have the same sequence, this information will be merged. This can reduce redundant data, improve analysis efficiency, and ensure the simplicity and accuracy of the results.

[0222] The step S510 preferably includes:

[0223] Step (1): Compare each low-frequency variation detection result. If there is the same mutation information and the same sequence, merge the two low-frequency variation detection results to achieve the purpose of de-redundancy.

[0224] Step (2): comparing the mutation information of the overlapping regions of the adjacent analysis windows. If the mutation information and sequence are identical, merging the mutation information of the two overlapping regions to achieve the purpose of removing redundancy.

[0225] The above step (2) will merge the mutation information in the overlapping area of ​​two adjacent windows similar to the previous step (1). The purpose of this step is to ensure that the overlapping area between windows does not lead to redundant records and to ensure the continuity and consistency of the mutation information.

[0226] The main purpose of the above steps is to remove redundancy. Removing redundancy can reduce the amount of data, improve computing efficiency, and ensure the simplicity and accuracy of the analysis results. This is especially important for the processing of large-scale genomic data.

[0227] In addition, by merging overlapping regions and removing redundancy, confusion caused by duplicate records can be reduced, ensuring that each mutation is recorded only once, thereby improving the reliability of the results.

[0228] Step S520 ensures that the mutation information finally outputted has a certain degree of credibility and reliability, and eliminates possible false positive results.

[0229] As mentioned above, the filtering criteria can be: if the ratio of reads supporting the variant is greater than 0.002, then the number of reads is greater than 2. Through strict filtering criteria, it is ensured that the final output result is high-quality and reliable. This step can remove low-confidence mutation information (such as false positives) and improve the accuracy of the test results.

[0230] In short, the processing of these steps can effectively reduce redundant information, improve the efficiency of data processing and the reliability of results, which is very important for the accurate detection of low-frequency variants.

[0231] Example:

[0232] In this example, 20 mitochondrial low-frequency mutations were simulated based on the human reference genome hg19, and a horizontal comparison test was performed.

[0233] Experimental methods:

[0234] (1) Conventional method (deepvariant);

[0235] (2) The method of the present invention (grandSNP);

[0236] The above two methods and corresponding software systems were used to identify mutations.

[0237] Experimental results:

[0238] Referring to Table 5, the following results were obtained through the above method and the corresponding software system:

[0239] Table 5. Horizontal test results

[0240] software All mutations The real mutation Deepvariant (conventional technique) 27 0 grandSNP (present invention) 4180 13

[0241] (1) Among conventional techniques, Deepvariant identified a relatively small number of mutations (27), but failed to detect any simulated real mutations, which indicates that this method has limitations in detecting low-frequency mutations and is not accurate enough.

[0242] (2) Using the method of the present invention and the corresponding grandSNP system, a large number of mutations (4180) can be identified, and 13 real simulated mutations can be identified, which shows that the method has a high sensitivity in detecting low-frequency mutations.

[0243] In addition, reference Fig.12 This embodiment also provides a low-frequency single nucleotide variation detection device, comprising:

[0244] A data comparison module 10 is used to obtain sequencing data and a corresponding reference genome, and compare the sequencing data to the reference genome to obtain comparison information;

[0245] A site extraction module 20, used for extracting potential variation positions as candidate sites based on the comparison information;

[0246] A site coverage module 30 is used to extract sequence information covering the candidate site as a candidate sequence;

[0247] A multiple alignment module 40 performs a multiple sequence base alignment optimization process on the candidate sequences to obtain an optimized alignment result;

[0248] The result output module 50 is used to output the low-frequency variation detection result based on the optimized comparison result.

[0249] It should be noted that the data comparison module in the low-frequency single nucleotide variation detection device of the present invention is used to obtain sequencing data and the corresponding reference genome, and compare the sequencing data to the reference genome to obtain comparison information. The way in which this module obtains sequencing data can refer to the prior art and depend on the situation. For example, the sequencing data stored in the sequencing device can be directly connected and called; or the sequencing sequence stored in another storage device or web page can be accessed and called through the network; or the sequencing data stored in an external storage device, such as a USB flash drive, a mobile hard disk, etc. can be accessed and called; or a storage module can be added to the device of the invention according to demand, and the sequencing data is stored in the storage module, and the data comparison module directly accesses the storage module to obtain the sequencing data. According to the product design, the data comparison module can also temporarily store or long-term store the sequencing data it calls.

[0250] In addition, the present invention also provides a low-frequency single nucleotide variation detection system, comprising a memory and a processor, wherein the memory stores a low-frequency single nucleotide variation detection program, and the processor runs the low-frequency single nucleotide variation detection program to enable the low-frequency single nucleotide variation detection system to perform the low-frequency single nucleotide variation detection method as described above.

[0251] In addition, the present invention also provides a computer-readable storage medium, on which a low-frequency single nucleotide variation detection program is stored. When the low-frequency single nucleotide variation detection program is executed by a processor, the low-frequency single nucleotide variation detection method as described above is implemented.

[0252] In summary, the low-frequency single nucleotide variation detection method provided by the present invention significantly improves the accuracy and sensitivity of the detection by improving the alignment and analysis process. Through multiple sequence alignment and re-alignment, false positive results caused by sequencing errors are reduced, thereby improving the accuracy of the results; there is no minimum requirement for the number of reads supporting mutations, thereby maintaining the sensitivity of low-frequency mutation identification and not missing mutations; at the same time, stacking analysis is used to comprehensively identify low-frequency variation sites, which is particularly suitable for the processing of third-generation sequencing data. This method can effectively distinguish true variations from pseudo-variations, reduce the false positive rate, and improve the reliability of the detection results.

[0253] The method provided by the present invention is widely applicable to third-generation sequencing data. This method performs well in applications requiring high-sensitivity detection, and the systematic and automated process improves detection efficiency, providing more reliable and efficient technical support for genomic research and clinical applications.

[0254] Through the description of the above implementation scheme, those skilled in the art can clearly understand that the above-mentioned embodiment method can be implemented by means of software plus the necessary general hardware platform, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention can be executed as a computer software product, which is stored in a storage medium (such as ROM / RAM, disk, CD) as described above, including a number of instructions to enable a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present invention. The above is only a preferred embodiment of the present invention, and the scope of the present invention is not limited thereto. All equivalent structures or equivalent process transformations made using the contents of this specification and the accompanying drawings, or directly or indirectly used in other related technical fields, are similarly included in the scope of protection claimed by the present invention.

Claims

1. A method for detecting low-frequency single nucleotide variation, characterized in that: include: S100, acquiring sequencing data and a corresponding reference genome, and aligning the sequencing data to the reference genome to obtain alignment information; S200, extracting potential variation positions as candidate sites based on the comparison information; S300, extracting sequence information covering the candidate site as a candidate sequence; S400, performing position alignment on the same parts of bases between the candidate sequences to obtain a multiple sequence alignment result; and removing the same gaps from the sequences in the multiple sequence alignment result according to the coordinate positions, and repeatedly performing the alignment process of bases in different directions until the results of two adjacent alignment processes are unchanged, thereby obtaining an optimized alignment result; S500: Outputting a low-frequency variation detection result based on the optimized comparison result.

2. The method for detecting low-frequency single nucleotide variation according to claim 1, characterized in that: The S100 includes: S110, obtaining the sequencing data and the reference genome, performing comparison analysis on the sequencing data based on the reference genome, and obtaining a comparison result file containing the location of the sequencing data in the reference genome; wherein the comparison result file is a file after sorting and indexing; S120, performing stacking and statistical processing on the comparison result file to obtain the comparison information.

3. The method for detecting low-frequency single nucleotide variation according to claim 2, characterized in that: The S120 includes: S121, dividing the reference genome into multiple analysis windows; S122, performing cigar character string analysis on each base of the sequencing data on the analysis window to obtain an analysis result; wherein the alignment event type in the analysis result includes at least one of a homomatch, a mismatch, an insertion, and a deletion; S123, using the analysis result as the comparison information based on the base position corresponding to each base.

4. The method for detecting low-frequency single nucleotide variation according to claim 2, characterized in that: The S200 includes: S210, counting the number of the alignment event types corresponding to each base position; S220, determining whether the number of the alignment event types is not less than 2, and whether the number of reads supported by each alignment event type is not less than the base position of the preset filtering standard; S230, if yes, the base position is a potential variation position, and the base position is used as the candidate site; S240, if not, the base position is not a potential variation position and is not used as the candidate site; In S220, the method for determining the preset filtering standard includes: S221, obtaining the sequencing depth of the current region; S222, if the sequencing depth is not less than 10, the preset filtering criteria is 2; S223, if the sequencing depth is less than 10, the preset filtering standard is 1; The S200 further includes: S600, determining the position interval between two adjacent candidate sites; S700: Merge two adjacent candidate sites whose position interval is less than 2 into one candidate site.

5. The method for detecting low-frequency single nucleotide variation according to claim 1, characterized in that: The S300 includes: S310, obtaining the coordinate position of the candidate site; S320, taking the coordinate position as the center, extracting the base corresponding to the coordinate position and the adjacent bases with a preset number of interval positions before and after to form a sequence covering the candidate site, and using the sequence as the candidate sequence.

6. The method for detecting low-frequency single nucleotide variation according to claim 1, characterized in that: The S400 includes: S410, comparing the candidate sequences in pairs to determine the identical partial bases in the sequences as identical components; S420, using the same components as a reference, using gaps to fill in unmatched positions of the candidate sequences, so that the candidate sequences are aligned through the same components to obtain the multiple sequence alignment result; wherein the total length of the bases of each sequence in the multiple sequence alignment result is equal to the total length of the gaps; S430, deleting gaps at the same position between sequences in the multiple sequence alignment result; S440, performing alignment processing of bases in different directions to obtain a first optimization result; S450, performing the alignment process of the bases in different directions on the first optimization result again to obtain a second optimization result; S460, determining whether the second optimization result is consistent with the first optimization result; S470: If yes, use the first optimization result or the second optimization result as the optimization comparison result; S480: If not, the second optimization result is returned to execute S450.

7. The method for detecting low-frequency single nucleotide variation according to claim 6, characterized in that: The alignment process of the bases in different directions in S450 includes: S451, performing left alignment processing and right alignment processing on each candidate sequence to obtain an arrangement result; S452, deleting gaps at the same position between sequences in the sorting result; S453, repairing the gaps at the ends of each candidate sequence.

8. The method for detecting low-frequency single nucleotide variation according to claim 7, characterized in that: The left alignment process includes: In the multiple sequence alignment result, a gap in a candidate sequence of the multiple sequence alignment result is determined; and the candidate sequence is used as a first original sequence, and the base adjacent to the right of the gap in the first original sequence is used as a right base; another candidate sequence in the multiple sequence alignment result to be compared is used as a first comparison sequence, and the base corresponding to the position of the gap in the first comparison sequence is used as a first comparison base; If the right base is the same as the first comparison base, the position of the right base in the first original sequence is swapped with the gap so that the position of the right base in the first original sequence is the same as the first comparison base in the first comparison sequence and the bases are the same; The right alignment process includes: In the multiple sequence alignment result, a gap in a candidate sequence of the multiple sequence alignment result is determined; and the candidate sequence is used as a second original sequence, and the base adjacent to the left of the gap in the second original sequence is used as a left base; another candidate sequence in the multiple sequence alignment result to be compared is used as a second comparison sequence, and the base corresponding to the position of the gap in the second comparison sequence is used as a second comparison base; If the left base is the same as the second comparison base, the position of the left base in the second original sequence is swapped with the gap so that the position of the left base in the first original sequence is the same as the second comparison base in the second comparison sequence and the base is the same.

9. The method for detecting low-frequency single nucleotide variation according to claim 1, characterized in that: The S500 includes: S510, taking the mutated bases in the sequence of the optimized alignment result as mutation information, and merging and removing redundancy of the mutation information; S520, filtering the mutation information after the combined de-redundancy processing according to a preset filtering standard to obtain the low-frequency mutation detection result.

10. A low-frequency single nucleotide variation detection device, characterized in that: include: A data comparison module, used to obtain sequencing data and a corresponding reference genome, and compare the sequencing data to the reference genome to obtain comparison information; A site extraction module, used for extracting potential variation positions as candidate sites based on the alignment information; A site coverage module, used to extract sequence information covering the candidate site as a candidate sequence; A multiple alignment module is used to align the same parts of the bases between the candidate sequences to obtain a multiple sequence alignment result; and remove the same gaps in the sequences in the multiple sequence alignment result according to the coordinate positions, and repeatedly align the bases in different directions until the results of two adjacent alignment processes are unchanged, thereby obtaining an optimized alignment result; A result output module is used to output a low-frequency variation detection result based on the optimized comparison result.

Images