Method and apparatus for aligning and correcting genetic sequence information

Abstract

The present invention aims to provide a method and apparatus for gene sequence information comparison and correction, comprising: inputting data to be sequenced for third-generation sequencing; performing a coarse comparison on the data using a half-and-half comparison strategy to obtain coarsely compared data; parsing the coarsely compared data to obtain parsed data; constructing a comparison database; and comparing the parsed data with the comparison database. BLASN Detailed alignment yields the aligned data; the sequence to be compared is set. query And compare any of the finely aligned data with the compared sequence. query Alignment; the data after fine alignment with the sequence being compared. query query During the alignment process, a new base is generated by voting on the bases at any position, and the problem of the current sequence is determined by the new base; this disclosure has good correction capability and time efficiency.

Description

Technical Field

[0001] This invention belongs to the field of bioinformatics, and in particular relates to a method and apparatus for gene sequence information comparison and correction. Background Technology

[0002] Currently, mainstream gene sequence alignment tools (such as minimap2 and BLASTN) typically employ a core strategy of "first selecting seeds, then extending the alignment." This method first selects several seed fragments in the sequence as initial matching points, and then expands outwards from these seeds to complete the alignment of the entire sequence. However, in highly repetitive or structurally complex genomic regions, such methods may miss some homologous matches or fail to identify the optimal match due to the randomness or insufficient coverage of seed selection. For example, although minimap2 calculates matching scores using multiple seed positions, in regions with dense repetitive sequences, the random distribution of seeds may not fully represent all potential matches, leading to missed matches or matching scores that are not the highest globally.

[0003] Therefore, this invention proposes a two-stage gene sequence alignment and correction method. This method first performs similarity filtering at the full-length level, and then conducts in-depth analysis of each local alignment result. It balances accuracy with overall sequence alignment speed. Summary of the Invention

[0004] The purpose of this invention is to provide a method and apparatus for gene sequence information alignment and correction, particularly for high-precision, long-read sequencing data generated by third-generation sequencing technology. This invention proposes a method to solve the difficulty of aligning and correcting highly repetitive sequences in third-generation sequencing, thereby addressing at least one technical problem in the prior art.

[0005] The technical solution of this invention is:

[0006] A method for gene sequence information alignment and correction includes:

[0007] Input the data to be sequenced for third-generation sequencing;

[0008] The data is coarsely compared using a half-and-half comparison strategy to obtain the coarsely compared data.

[0009] The coarsely compared data is parsed to obtain the parsed data; and a comparison database is constructed.

[0010] The parsed data is compared with the comparison database using a fine BLASN comparison to obtain the finely compared data.

[0011] Set the comparison sequence query, and align any of the finely aligned data with the comparison sequence query;

[0012] During the process of aligning the data after fine alignment with the query sequence being compared, a new base is generated by voting on the bases at any position, and the problem of the current sequence is determined by the new base.

[0013] The method of using a half-and-half comparison strategy to perform a coarse comparison on the data to obtain coarsely compared data includes:

[0014] The data to be sequenced for the third generation is parsed in fa file format and standardized; the sequence information in the fa file is obtained and the minimizer data of any sequence is obtained; the minimizer data includes: the minimizer data of the first half of the sequence, the minimizer data of the second half of the sequence, and the overall minimizer data obtained by dividing the minimizer data equally according to the sequence length.

[0015] The similarity between the minimizers data of any sequence and the minimizers information of all other sequences is calculated one by one to obtain the calculation results;

[0016] The calculation results are filtered using a preset similarity threshold to obtain data after coarse comparison;

[0017] The total minimizer information includes the sum of minimizer data for any sequence.

[0018] The similarity calculation includes:

[0019] ;

[0020] in, The similarity between the sequences being compared and the sequences being compared. The number of minimizers participating in the comparison; The total number of minimumizers for the aligned sequence; This refers to the number of times common minimizers appear in both the alignment sequence and the aligned sequence. Common minimizers are those present in both sequences.

[0021] The process of parsing the coarsely compared data to obtain parsed data and constructing a comparison database includes:

[0022] Obtain the data after coarse comparison, including the output file rough.csv after coarse comparison;

[0023] Parse the output file rough.csv to obtain the seq_id column of the aligned sequence index, the pre_id column of the index of the first half of the aligned sequence length, the last_id column of the index of the second half of the aligned sequence length, and the all_id column of the index of the overall length of the aligned sequence.

[0024] Based on the seq_id column of the aligned sequence index number, obtain the pre_id column of the first half of the aligned sequence length, the last_id column of the second half of the aligned sequence length, and the all_id column of the overall length of the aligned sequence corresponding to any aligned sequence index number seq_id column. Construct a set of pre_id columns of the first half of the aligned sequence length, a set of last_id columns of the second half of the aligned sequence length, and a set of all_id columns of the overall length of the aligned sequence.

[0025] For each ID appearing in the pre_id column of the first half of the alignment sequence, take the first half of its sequence; for each ID appearing in the last_id column of the second half of the alignment sequence, take the second half of its sequence; and for each ID appearing in the all_id column of the overall length of the alignment sequence, take the entire sequence, and construct an alignment database.

[0026] The process of performing a fine-grained BLASN comparison between the parsed data and the comparison database to obtain the finely compared data includes:

[0027] The data obtained after coarse comparison is parsed to obtain the seq_id column of the compared sequence index number, the pre_id column of the index number of the first half of the compared sequence length, the last_id column of the index number of the second half of the compared sequence length, and the all_id column of the index number of the overall length of the compared sequence.

[0028] The seq_id column of the sequence being compared is subjected to a fine BLASN comparison with the comparison database to obtain the fine comparison result.

[0029] Based on the bitscore value and match length in the fine comparison results, obtain the relative match quality score:

[0030] ; ;

[0031] Where Rms is the relative matching quality score; The bit score is the result of normalizing the original alignment score. S is the scale parameter, which is a scoring matrix composed of the number of mismatches, the total number of gaps, and the gap length. S is the original score, and K is the statistical parameter. The length of the sequence to be matched.

[0032] The sequence for setting the seq_id index is the alignment sequence, referred to as the alignment sequence query. Any of the alignment sequences, typically called the subject sequence, is aligned with the alignment sequence query, including:

[0033] Select the n smallest sequences from pre_id, last_id, and all_id. Since pre_id, last_id, and all_id point to different parts of the same sequence, obtain the relative matching quality score Rms value from the index columns of the first half of the alignment sequence (pre_id), the second half (last_id), and the overall length of the alignment sequence (all_id) in the subject sequence. Sort the sequences by relative matching quality score Rms value from largest to smallest, and select the top n as the sequences to be processed (n is a natural number).

[0034] During the sorting process, if the ids are the same, only the sequence portion corresponding to the highest Rrms value is retained; the id includes: the index number pre_id column for the first half of the sequence length, the index number last_id column for the second half of the sequence length, and the index number all_id column for the overall length of the sequence length.

[0035] Delete the gap in the compared sequence query and record the information of the deleted gap to obtain the compared sequence query without gap;

[0036] Using the gap-free alignment sequence query, the subject sequence is sequentially aligned with the start and end positions in BLASTN.

[0037] The process of aligning the data after fine alignment with the query sequence being compared involves voting on bases at any position to generate new bases, and using these new bases to determine problems with the current sequence, including:

[0038] choose The top 15 sequences enter adaptive position voting, and the number of times each position in any sequence aligns with the compared sequence query is used as the number of votes to participate in the voting;

[0039] When the number of alignments reaches the voting threshold, the voting process begins.

[0040] When the number of alignments does not reach the voting threshold:

[0041] Sequentially obtain the number of votes for each of the remaining sequences until the voting threshold is reached; or,

[0042] If the number of votes for any remaining sequence is less than the voting threshold, then voting is performed first. The adaptive positional voting rule is as follows:

[0043] In the query sequence, consecutive and identical bases at any position constitute the same base string. If the number of identical bases in the aligned subject sequence is greater than the minimum confidence level, then... If the sequence is correct, the vote is successful; otherwise, the bases in the compared sequence query are retained as new bases. If the lengths of the two identical base strings are inconsistent, it indicates that there is a gap caused by sequencing errors, and a gap placeholder is added at the beginning of the string.

[0044] After all positions in the compared sequence query have voted, a new query sequence is obtained;

[0045] Perform a BLASTN operation on the new query sequence and the original comparison sequence query to obtain the alignment result of the two sequences, and save the alignment result of the two sequences; this can be used to detect whether the current correction result is reliable.

[0046] The alignment results of the two sequences include: the number of mismatches and / or gaps between the two sequences.

[0047] An electronic device for gene sequence alignment and correction, comprising:

[0048] Storage media, used to store computer programs

[0049] The processing unit exchanges data with the storage medium and executes the computer program to perform the steps of the gene sequence information comparison and correction method as described above when performing gene sequence information comparison and correction.

[0050] A computer-readable storage medium:

[0051] The computer-readable storage medium stores a computer program.

[0052] When the computer program is running, it performs the steps of the gene sequence information comparison and correction method described above.

[0053] The beneficial effects of the present invention include at least the following:

[0054] The gene sequence alignment and correction method disclosed in this invention is based on a two-stage gene sequence alignment process: coarse alignment and fine alignment. It reduces the computation time of coarse alignment by calculating a minimizer and uses only highly similar sequences for the second stage of fine alignment. During fine alignment, the BLASTN tool is used to generate new bases by voting on each position in the query of the sequence being aligned. Finally, the new bases are used to determine whether the current sequence contains accidental sequencing errors or other problems, demonstrating good correction capability and time efficiency. Attached Figure Description

[0055] Figure 1 This is a flowchart of the gene sequence alignment and correction method described in this application;

[0056] Figure 2 This is a statistical chart showing the correction results of the gene sequence alignment and correction method described in this application;

[0057] Figure 3 A graph showing the number of subjects voting for each base position in a given comparison sequence query;

[0058] Figure 4 This is a schematic diagram showing the alignment of the sequences of subject1 and subject2 with the comparison sequence query. Detailed Implementation

[0059] The present application will now be further described with reference to the accompanying drawings. Specific Implementation Example 1:

[0061] This disclosure provides an embodiment:

[0062] This invention proposes a gene sequence alignment and correction method based on two processes: coarse alignment and fine alignment. It aims to solve the problem of the difficulty in accurately matching and correcting highly repetitive sequences in third-generation sequencing data.

[0063] The technical solution of this embodiment is implemented in two stages: a coarse alignment stage and a fine alignment stage. The gene sequence alignment and correction method described in this embodiment is developed and implemented using Python 3.10, and the implementation block diagram is as follows. Figure 1 As shown. In this embodiment, the columns for the index number seq_id of the aligned sequence, the index number pre_id of the first half of the aligned sequence, the index number last_id of the second half of the aligned sequence, and the index number all_id of the overall length of the aligned sequence are simplified to: seq_id, pre_id, last_id, and all_id.

[0064] In the rough alignment stage: the input is a third-generation sequencing file, supporting fa and fa.gz formats, and the output is a rough.csv file (default). Customizable parameters during this process include: input file location (input), output file location (output), the number of minimizer processes (nc), k-mer length (k), k-mer window size (w), similarity threshold (kt), and the number of sequence alignment processes (np).

[0065] Specifically, the detailed process of the coarse comparison is as follows:

[0066] First, the input file is standardized and parsed into FA file format to obtain the FA file. Then, a producer-consumer model is initiated. Producers are responsible for reading the sequence information from the FA file, while consumers are responsible for calculating the minimizers for the first half, second half, and the entire sequence. The number of producers is controlled by the parameter `nc`. After obtaining the minimizer statistics for the three parts, the similarity between the three parts of each sequence and the complete minimizers of other sequences is calculated.

[0067] The formula for calculating similarity is:

[0068] ;

[0069] in, The number of minimizers participating in the comparison; The number of full-length minimizers for the aligned sequence; This refers to the number of times the common minimizers of the alignment sequence and the alignment sequence occur; This refers to the similarity between the sequence being compared and the sequence being compared.

[0070] The similarity threshold is pre-specified by the parameter kt, with a default value of 0.70. When the similarity exceeds the threshold, the sequences involved in the alignment will be saved to the CSV file specified by output. Figure 1 The rough.csv file in the file.

[0071] In this embodiment, the split-half alignment strategy avoids the problem that only the first or second half of the subject matches the query sequence highly, yet fails to enter the precise alignment stage. For example... Figure 4 As shown, for sequences subject1 and subject2, if the split-half strategy is not used, the similarity value will be calculated... It is difficult to reach the threshold of 0.70 and be filtered out.

[0072] Specifically, the detailed comparison phase includes the following steps:

[0073] First, the output file rough.csv of the coarse alignment is parsed. The file contains seq_id, pre_id, last_id, and all_id. Then, seq_id is iterated, and the pre_id, last_id, and all_id corresponding to each seq_id are retrieved. For IDs appearing in the pre_id set, the first half of their sequence is taken; for IDs appearing in the last_id set, the second half of their sequence is taken; and for IDs appearing in the all_id set, the entire sequence is taken. This constructs the BLASTN alignment database.

[0074] Furthermore, a fine-grained BLAS NMR alignment is performed between the seq_id and the alignment database, filtering out sequences shorter than 5000bp from the alignment results. Since the pre_id, last_id, and all_id may contain the same ID, but only different parts are used for the alignment, only these parts are considered. The sequence portion with the largest value is included in the subsequent correction process, i.e., deduplication.

[0075] The calculation formula is:

[0076] ;

[0077] ;

[0078] Where Rms is the relative mismatch rate matching quality score; bitscore is the bit score, which is the result after normalizing the original alignment score; λ is the scaling parameter, which is a scoring matrix composed of the number of mismatches, the number of gaps, and the gap length; S is the original score; and K is the statistical parameter. To compare and match the sequence length. In this embodiment, Taking into account indicators such as similarity, mismatch rate, gap degree, and information density per unit length, the regions with the highest matching quality were selected to enter the subsequent voting stage.

[0079] Then, select from the deduplicated pre_id, last_id, and all_id. The 50 largest sequences are selected and aligned with the query sequence being compared. However, gaps exist in the query, making alignment impossible. Therefore, gaps in the query need to be removed and recorded. Queries without gaps can be aligned with the subject sequentially based on the start and end positions in BLASTN.

[0080] Choose again The first 15 sequences are used for voting. The number of alignments at each position is calculated, which is the number of positions eligible for voting. This number is between 0 and 15. Generally, the number of alignments at the beginning and end is relatively smaller. When the number of alignments reaches 10, voting begins. If the number is insufficient, usually at the beginning or end, sequences are taken from the remaining 35 sequences until the minimum number of positions eligible for voting (10) is reached before voting. If all 35 sequences have been taken and the number of positions eligible for voting is still less than 10, then voting proceeds first. If the same base is greater than... (Default value 5) If the vote is successful, the bases in the query remain the new bases. Once all positions in the query sequence have been voted on, a new query sequence is generated, along with the corresponding mismatch number, gap number, etc. The new sequence and the original query are then subjected to another BLASTN operation to obtain the alignment results, which are saved in the `accurate.csv` file. Intermediate results, such as the sequences participating in the voting and the mismatch positions, are saved in `ts.csv`. Finally... Figure 2 The number of mismatches, gaps, and gap_opens in the sequence is visualized. Figure 3 The table lists the number of subjects participating in the voting for each base position of a given query sequence.

[0081] This embodiment discloses a gene sequence alignment and correction method based on two processes: coarse alignment and fine alignment. It reduces the computation time of coarse alignment by calculating a minimizer and uses only highly similar sequences for the second stage of fine alignment. During fine alignment, the BLASTN tool is used to generate new bases by voting on the bases at each position in the sequence query. These new bases are then used to determine whether the current sequence contains accidental sequencing errors or other problems.

[0082] Verification process:

[0083] Using the gene sequence alignment and correction method described in Specific Example 1, PacBio HiFi third-generation sequencing data of porcine chromosomes 3 and 10 were tested. Approximately 1.5GB of fa files underwent coarse and fine alignment using 50 processes, with a total time of about 10 hours. The generated corrected new sequences were validated for assembly quality using the existing Yak tool, with a final QV value of 40.13. The number of corrected bases and alignment accuracy both exceeded those of the minimap2 tool. Based on the above, it is evident that the gene sequence alignment and correction method provided in this embodiment has good correction capability and time efficiency. Specific Implementation Example 2:

[0085] The present invention also provides an embodiment:

[0086] An electronic device includes: a storage medium and a processing unit; wherein the storage medium is used to store a computer program, and the processing unit exchanges data with the storage medium for executing the computer program through the processing unit during gene sequence alignment and correction to perform the steps of the gene sequence alignment and correction method as described in Specific Embodiment 1.

[0087] A computer-readable storage medium storing a computer program; when the computer program is run, it performs the steps of the gene sequence alignment and correction method as described in Specific Embodiment 1.

[0088] In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wireline, optical fiber, RF, etc., or any suitable combination thereof.

[0089] The above descriptions only cover a few specific embodiments of the present invention. However, the present invention is not limited thereto, and any variations that can be conceived by those skilled in the art should fall within the protection scope of the present invention. The above-mentioned serial numbers are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

Claims

1. A method of aligning and correcting genetic sequence information, characterized by, include: Input the data to be sequenced for third-generation sequencing; The data is coarsely compared using a half-and-half comparison strategy to obtain the coarsely compared data. The coarsely compared data is parsed to obtain parsed data. Using the parsed data, three index sets are created by grouping the aligned sequence index numbers. Gene fragments corresponding to each index are then extracted, and all gene fragments are aggregated to complete the alignment database construction. The three index sets are: the set of index numbers for the first half of the aligned sequence, the set of index numbers for the second half of the aligned sequence, and the set of index numbers for the entire aligned sequence. The gene fragments include: the first half of the sequence, the second half of the sequence, and the complete sequence. The parsed data is compared with the comparison database. A fine comparison is performed to obtain a fine comparison result. The fine comparison result is then filtered by a preset relative matching quality score threshold to obtain the finely matched data. The relative matching quality score is obtained based on the bit score value and matching length in the fine comparison result. Set the alignment sequence and align any of the finely aligned data with the alignment sequence; During the process of aligning the data after fine alignment with the aligned sequence, a new base is generated by voting on the bases at any position, and the problem of the current sequence is determined by the new bases; the coarse alignment is performed on the data using a half-and-half comparison strategy to obtain the coarsely aligned data, including: The data to be sequenced for the third generation are as follows: File format parsing and standardization processing; obtaining... The sequence information in the file and the acquisition of any sequence Data; any sequence The data includes: based on sequence length The data is divided equally, and the first half of the sequence is obtained. Data, the latter half Data and overall data; Any sequence The data are completely different from all other sequences. The information is similar to each other, and the results are obtained. The calculation results are filtered using a preset similarity threshold to obtain data after coarse comparison; The similarity calculation includes: ； in, The similarity between the sequences being compared and the sequences being compared. For those participating in the comparison quantity; For the entire alignment sequence quantity; For the common features in the alignment sequence and the alignment target sequence Number of times it appears.

2. The gene sequence information comparison and correction method according to claim 1, characterized in that, The data after coarse comparison is parsed to obtain parsed data; And construct a comparison database, including: Obtain the data after coarse comparison, including the output file after coarse comparison; Parse the output file to obtain the index number column of the aligned sequence, the index number column of the first half of the aligned sequence, the index number column of the second half of the aligned sequence, and the index number column of the overall length of the aligned sequence; Based on the index number column of the alignment sequence, obtain the index number column of the first half of the alignment sequence, the index number column of the second half of the alignment sequence, and the index number column of the overall length of the alignment sequence corresponding to any alignment sequence index number column. Construct a set of index number columns of the first half of the alignment sequence, a set of index number columns of the second half of the alignment sequence, and a set of index number columns of the overall length of the alignment sequence. For each occurrence of an index number in the first half of the alignment sequence, take the first half of its sequence; for each occurrence of an index number in the second half of the alignment sequence, take the second half of its sequence; and for each occurrence of an index number in the overall length of the alignment sequence, take the entire sequence, to construct an alignment database.

3. The gene sequence information comparison and correction method according to claim 1, characterized in that, The process involves performing a fine comparison between the parsed data and the comparison database to obtain a fine comparison result. This fine comparison result is then filtered using a preset relative matching quality score threshold to obtain finely compared data, including: The data obtained after coarse comparison is parsed to obtain the index number column of the compared sequence, the index number column of the first half of the length of the compared sequence, the index number column of the second half of the length of the compared sequence, and the index number column of the overall length of the compared sequence. The index number column of the sequence being compared is compared with the comparison database to obtain a detailed comparison result. The relative matching quality score is obtained based on the bit score and matching length in the fine comparison results.

4. The gene sequence information comparison and correction method according to claim 3, characterized in that, include: The relative matching quality score is obtained using the following formula: ； ； in, This refers to the relative quality score of the match. Bit fraction; For scale parameters; This is the raw score; These are statistical parameters; The length of the sequence to be matched.

5. The gene sequence information comparison and correction method according to claim 1, characterized in that, The step of setting the alignment sequence and aligning any of the finely aligned data with the alignment sequence includes: The sequence whose index is being compared is called the alignment sequence. ; and any of the alignment sequences described herein shall be referred to as sequence; Obtain the respectively The relative matching quality scores of the index number column for the first half of the alignment sequence, the index number column for the second half of the alignment sequence, and the index number column for the overall length of the alignment sequence; Based on relative matching quality score Sort values ​​from largest to smallest and take the first few. The strips are the sequences to be processed. It is a natural number; Delete the compared sequence In And record deletion Information obtained The sequence being compared ; Using the aforementioned The sequence being compared according to The start and end positions in will The sequence is aligned with it sequentially; During the sorting process, If they are the same, only the highest value is retained. The sequence portion corresponding to the value; the aforementioned This includes: the index number column for the first half of the alignment sequence, the index number column for the second half of the alignment sequence, and the index number column for the overall length of the alignment sequence.

6. The gene sequence information comparison and correction method according to claim 5, characterized in that, The process of aligning the data after fine alignment with the aligned sequence, generating new bases by voting on bases at any position, includes: Select the one with the larger relative matching quality score. Each sequence enters an adaptive positional voting process, and the number of times each position in any sequence aligns with the compared sequence is used as the number of votes to participate in the voting. It is a natural number; When the number of alignments reaches the voting threshold, the voting process begins. When the number of alignments does not reach the voting threshold: Sequentially obtain the number of votes for each of the remaining sequences until the voting threshold is reached; or, If the number of votes for any remaining sequence is less than the voting threshold, then voting is performed first. The adaptive positional voting rule is as follows: Aligned sequence Consecutive and identical bases at any position in the sequence constitute the same base string, if aligned If the number of identical bases in the sequence is greater than the minimum confidence level, the vote is successful; otherwise, the compared sequence is maintained. The bases in the sequence are used as new bases; if the two are the same but the lengths of the base strings are inconsistent, it indicates a sequencing error. Add at the beginning of the string Placeholder; When the sequence is compared After voting in all positions, a new result is obtained. sequence; New Sequence and the original aligned sequence conduct The operation is performed to obtain the alignment results of the two sequences, and the alignment results of the two sequences are saved. The alignment results of the two sequences include: the number of mismatches and / or gaps between the two sequences.

7. An electronic device for gene sequence alignment and correction, characterized in that, include: Storage media, used to store computer programs The processing unit exchanges data with the storage medium and executes the computer program to perform the steps of the gene sequence information comparison and correction method as described in any one of claims 1-6 when performing gene sequence information comparison and correction.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program. When the computer program is run, it performs the steps of the gene sequence information comparison and correction method as described in any one of claims 1-6.

Images