A DNA storage method with pseudo-noise sequence encoding

Through the pseudo noise sequence accompanying coding method, the problems of low data readout efficiency and poor reliability in DNA storage media are solved, and fast error-free data recovery under low coverage is achieved, reducing processing complexity and cost.

CN116707541BActive Publication Date: 2024-11-19TIANJIN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202310626234.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-05-30
Publication Date
2024-11-19
Estimated Expiration
2043-05-30

AI Technical Summary

Technical Problem

Existing DNA storage media have problems in long-term storage of low data read efficiency, poor reliability and high cost, especially due to the high insertion/deletion error rate, which requires high coverage of data recovery, which increases processing complexity and cost. .

Method used

Using the pseudo-noise sequence accompanying coding method, the information sequence is encoded through grouping error correction code, and a two-layer coding sequence is constructed with the pseudo-noise sequence, mapped into data DNA sequences, synthesized medium-length DNA fragments, and data is realized through efficient library construction and three-generation sequencing. Read out, use pseudo noise sequence alignment to identify and correct insertion/deletion errors to reduce processing complexity.

Benefits of technology

Fast and error-free recovery of data at low coverage is achieved. The pseudo noise sequence accompanied by encoding is highly robust to insert/delete errors, avoiding complex assembly operations, and reducing processing complexity and cost.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116707541B_ABST
    Figure CN116707541B_ABST
Patent Text Reader

Abstract

The present invention discloses a DNA storage method with pseudo-noise sequence accompanying coding. The method uses a group error correction code to encode an information sequence to generate a codeword sequence, and constructs a two-layer coding sequence with the pseudo-noise sequence, and then maps it to a data DNA sequence; synthesizes and assembles the data DNA sequence into a medium-length DNA as a storage medium; when the data is read out, the medium-length DNA is efficiently built and sequenced by the third generation, thereby generating a sequencing read segment; the sequencing read segment is demapped to obtain a damaged pseudo-noise sequence, which is compared with a known pseudo-noise sequence to determine the alignment position of the read segment and identify insertion / deletion errors, and then the codeword sequence is corrected and consensus and decoding are performed. The present invention has the ability to quickly recover data under low sequencing coverage, and its advantage is that the pseudo-noise sequence accompanying coding has high robustness to insertion / deletion errors, and the pseudo-noise sequence comparison avoids complex assembly operations, thereby reducing processing complexity.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of DNA data storage, and in particular to a method for designing medium-length DNA with pseudo-noise sequence accompanying coding and a method for rapid reading. Background Art

[0002] With the continuous development of digital information technology, the total amount of global data has increased explosively. According to the forecast of International Data Corporation, the total amount of global data will reach 175ZB in 2025 and will further increase by an order of magnitude by the end of this century. Therefore, the demand for storage media with higher density and longer service life is also increasing. The maximum density of existing storage media (CD-ROM, flash memory, tape, etc.) is 10 3 GB / mm 3 , and has reached its density limit, making it unsuitable for long-term storage. This means that if information needs to be stored for a long time, it needs to be regularly transferred to newer storage media, which has high information transfer costs and may face a large amount of data loss. For long-term information storage, an efficient storage medium is needed to circumvent this laborious and expensive process and effectively address other limitations associated with current storage media (including energy consumption and insufficient data density).

[0003] DNA, as a promising data storage medium, stores all genetic information in the form of four bases {A, T, C, G} to encode, transmit and preserve information. Compared with traditional storage media, DNA molecules can be stable for thousands of years under appropriate storage conditions. For example, 300,000-year-old mitochondrial DNA from a bear has been successfully sequenced. This DNA sample was preserved in the bones, thus proving that the power consumption required for DNA archival storage is very low. In addition to its stability, ease of synthesis and low storage cost, DNA has a major key advantage: data density. Based on its physical size, the theoretical data density of a DNA molecule is ~4.5×10 7 GB / g, which is about 7 orders of magnitude higher than the density achievable by traditional storage media currently in use.

[0004] However, the readout efficiency and reliability of data extracted from DNA are key issues. Traditional DNA storage media require biochemical reactions for data readout, which is inefficient. The third-generation nanopore sequencing can achieve rapid readout by quickly collecting the electrical signals generated when DNA molecules pass through nanoscale channels. However, the error rate of nanopore sequencing readout is relatively high, and the error types may include difficult-to-handle insertion / deletion errors. Insertion / deletion errors can lead to a large number of substitution errors, and a high sequencing coverage is required to achieve error-free data recovery, which has a serious impact on the performance of DNA data storage systems.

[0005] There are some methods for encoding and reading DNA, but they are costly, complex, and inefficient. For a pool of short oligonucleotides, Organick et al. used PCR and Gibson assembly technology to assemble multiple short oligonucleotide sequences into large DNA fragments. After nanopore sequencing, they successfully recovered a 32KB file at 36× sequencing coverage. This method can achieve third-generation nanopore sequencing for short DNA fragments, but complex sample operations are required at the data readout end, which increases cost overhead and reduces feasibility. At the same time, large DNA fragments can also be directly designed for nanopore sequencing readout, which ignores the complex DNA assembly operation and reduces the high writing cost. Yazdi et al. used constraint coding and homopolymer checksums to encode several images and constructed 17 DNA fragments of about 1kb in length. After nanopore sequencing, 200× sequencing coverage was required to achieve data recovery. Lopez et al. assembled a large DNA fragment of about 5kb from a pool of short oligonucleotides and successfully decoded 1.67MB of data at 22× sequencing coverage using nanopore sequencing. The researchers used an efficient low-density parity check (LDPC) code and a pseudo-random sequence superposition coding method to construct a 254kb yeast artificial chromosome (a very large DNA fragment) to store two pictures and a short video clip (total size 37.8KB). Using a nanopore sequencing instrument, fast and portable data recovery was achieved at a minimum sequencing coverage of 16.8×. Sun et al. designed a MEPCAL coding scheme to construct a DNA fragment of approximately 54kb to encode a 5.56KB text file, and the minimum sequencing coverage required for data recovery after nanopore sequencing was 9.13×. However, at the data read-out end, these methods require complex read assembly steps, which increases the complexity of processing.

[0006] In view of the above problems, the present invention proposes a medium-length DNA design method and a fast readout method accompanied by pseudo-noise sequence coding. After the information sequence is encoded by a group error correction code, a two-layer coding sequence is constructed with a pseudo-noise sequence, and the two-layer coding sequence is mapped to a data DNA sequence according to a specific mapping rule. The data DNA sequence is synthesized and assembled into a medium-length DNA fragment as a storage medium, and is subjected to efficient library construction and third-generation sequencing; when the data is read out, the generated sequencing reads are preprocessed, and the damaged pseudo-noise sequence obtained by demapping the reads close to the size of the DNA fragments is compared with the known pseudo-noise sequence to determine the read position and identify insertion / deletion errors, correct the damaged codewords, and perform consensus and decoding. The present invention realizes fast and error-free data recovery at low coverage, and its advantage is that the encoding method accompanied by the pseudo-noise sequence is highly robust to insertion / deletion errors; pseudo-noise sequence comparison avoids complex assembly operations and reduces processing complexity. Summary of the invention

[0007] The present invention provides a DNA storage method accompanied by pseudo-noise sequence encoding, which realizes fast and error-free data recovery at low coverage. The encoding method accompanied by pseudo-noise sequence is highly robust to insertion / deletion errors; pseudo-noise sequence comparison avoids complex assembly operations and reduces processing complexity, as described below:

[0008] A method for storing pseudo-noise sequence-accompanying coded DNA, the method comprising the following steps:

[0009] (1) Using a group error correction code to group encode the data to be stored by the user, generate a code word bit sequence, and generate a binary pseudo noise sequence with the same length as the code word bit sequence, divide the error correction coded code word sequence and the pseudo noise sequence into two layers, and construct two layers of coding sequences with exactly the same length, and map the two layers of coding sequences into data DNA sequences according to a specific bit pair to base mapping rule;

[0010] (2) Connecting the data DNA sequence with the vector sequence, synthesizing and assembling into different forms of DNA fragment entities using chemical synthesis or enzymatic synthesis, using the DNA fragments as storage media, and performing library construction and third-generation sequencing to generate sequencing reads;

[0011] (3) Demapping the sequencing reads into damaged codeword sequences and damaged pseudo-noise sequences, determining the alignment position of the reads and identifying the insertion / deletion errors therein by aligning the damaged pseudo-noise sequences with the known pseudo-noise sequences, correcting the damaged codeword sequences according to the insertion / deletion error positions, and performing a majority vote on the corrected codeword sequences bit by bit to obtain a consensus sequence, which is then sent to a decoder for decoding to restore the original data.

[0012] The method uses a group error correction code to group encode the data to be stored by the user, generates a code word bit sequence, and generates a binary pseudo noise sequence with the same length as the code word bit sequence, divides the error correction coded code word sequence and the pseudo noise sequence into two layers of bit-by-bit correspondence, constructs two layers of coding sequences with exactly the same length, and maps the two layers of coding sequences into data DNA sequences according to a specific bit pair to base mapping rule. The specific steps are as follows:

[0013] (1.1) A block error correction code (n, k) is used to encode an information sequence of length k into an error correction coded codeword sequence. The length of the coded codeword sequence is n bits. The block error correction code uses a low-density parity check code with parameters (n, k), a multi-level low-density parity check code defined on a high-order Galois field GF(q) (q>2), a low-density parity check code with a large girth, a Reed-Solomon code, etc., where k is the length of the information sequence, i.e., the number of information bits, and n is the length of the coded error correction codeword;

[0014] (1.2) The pseudo-random sequence accompanying the coded codeword bit by bit is obtained by periodically extending two m sequences of different orders to obtain two sequences of length n bits, and then performing bit-by-bit XOR on the sequences intercepted by the two periodic extensions of length n bits to obtain a pseudo-noise sequence of length n bits;

[0015] (1.3) constructing two-layer coding sequences from the error correction coding codeword sequence and the generated pseudo noise sequence of length n bits, taking out one bit from each layer one by one, and selecting one bit from each layer of coding sequence one by one to form a bit pair;

[0016] (1.4) Map the bit pairs to bases according to the mapping rules: 00→A, 01→T, 10→G, 11→C, and obtain a data DNA sequence of length n.

[0017] The data DNA sequence is synthesized and assembled into medium-length DNA fragments in different forms as storage media, and the DNA fragments are efficiently library-built and third-generation sequenced to generate sequencing reads. Specifically, the data DNA sequence is synthesized and assembled into medium-length DNA fragments in various forms such as circular plasmids or artificial chromosomes by means of methods such as in vivo assembly, and the medium is used as a storage medium. The present invention has high robustness to various types of base errors, and can use a third-generation sequencing platform with a high error rate (such as Nanopore and PacBio, etc.) for sequencing to generate sequencing reads.

[0018] The sequencing reads are preprocessed, the processed sequencing reads are demapped into damaged codeword sequences and damaged pseudo-noise sequences, the damaged pseudo-noise sequences are compared with known pseudo-noise sequences, the alignment positions of the reads are determined and insertion / deletion errors therein are identified, the damaged codeword sequences are corrected according to the insertion / deletion error positions, the corrected codeword sequences are subjected to majority voting bit by bit to obtain a consensus sequence, and the consensus sequence is sent to a decoder for decoding to restore the original data. The specific steps are as follows:

[0019] (4.1) Demap each base in the sequencing read into a bit pair according to the rule of A→00, T→01, G→10, C→11, where the upper bit sequence is the damaged codeword sequence and the lower bit sequence is the damaged pseudo noise sequence;

[0020] (4.2) The known pseudo-noise sequence of length n at the read end is extended to twice the length, and the damaged pseudo-noise sequence is compared with the extended twice-length sequence to obtain the optimal matching position of the damaged pseudo-noise sequence, obtain the alignment position of the sequencing read segment and identify the specific position of the insertion / deletion error of the damaged pseudo-noise sequence;

[0021] (4.3) Based on the fact that the positions where insertion and deletion errors occur in the damaged pseudo-noise sequence are exactly the same as the positions where insertion and deletion errors occur in the coded codeword, the damaged codeword sequence is corrected according to the positions of insertion / deletion errors in the damaged pseudo-noise sequence, that is, a mark is added to the position where the deletion error occurs, marked as an erasure error, represented by a symbol representing an erasure error, and the bit at the position where the insertion error occurs is directly removed, thereby correcting the insertion error and converting the deletion error into an erasure;

[0022] (4.4) The corrected codeword sequence is placed at the correct position obtained by comparison. Considering that some positions may be repeatedly overlapped by multiple codeword sequences, a majority decision is made on all the corrected codeword sequences at each position bit by bit to obtain a consistent sequence. The uncovered codeword parts are marked as erased, thus obtaining a complete block-coded codeword sequence, which is then sent to the decoder for decoding to restore the original data.

[0023] The modified codeword sequence is placed at the correct position obtained by comparison. Considering that some positions may have multiple codeword sequences that overlap repeatedly, a majority decision is made on all the modified codeword sequences at each position bit by bit to obtain a consistent sequence. The uncovered codeword parts are marked as erased, thereby obtaining a complete block-coded codeword sequence, which is then sent to a decoder for decoding to restore the original data. The specific steps are as follows:

[0024] (5.1) For multiple corrected codeword sequences, count the number of bits "1" or "0" in each position, select the value with the largest number of {0, 1} as the bit value at that position in the consistency sequence, and if there is no bit at that position or a majority vote cannot be made, mark it as an erasure error and represent it with a bit symbol representing an erasure error, and finally obtain a consistency sequence of length n;

[0025] (5.2) The consistency sequence is sent to the decoder of the error correction code for error correction including substitution and erasure to restore the original data. The decoder of the error correction code is matched with the encoding method, that is, the decoder of the low-density parity-check code with parameters (n, k), the multi-base low-density parity-check code defined on the high-order Galois field GF(q) (q>2), the low-density parity-check code with large girth, the Reed-Solomon code, etc.

[0026] The beneficial effects of the technical solution provided by the present invention are:

[0027] The pseudo-noise sequence-accompanying encoding DNA storage method proposed in the present invention, and the proposed high-reliability group error correction code and pseudo-noise sequence-accompanying encoding method encode the data file into the data DNA sequence; the data DNA sequence is synthesized and assembled into medium-length DNA fragments of different forms as DNA storage media; when the data is read out, the medium-length DNA is efficiently library-built and nanopore sequenced, and then the sequencing reads are pre-processed, and the damaged pseudo-noise sequence obtained by demapping the reads close to the length of the DNA fragments is compared with the known pseudo-noise sequence to determine the alignment position of the sequencing reads and identify the insertion / deletion errors therein, correct the damaged codewords and perform consensus and decoding. The present invention realizes fast and error-free data recovery under low coverage, and is characterized in that the encoding method accompanied by the pseudo-noise sequence has high robustness to insertion / deletion errors; the pseudo-noise sequence comparison avoids complex assembly operations and reduces processing complexity. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 A system block diagram for implementing the solution of the present invention;

[0029] Figure 2 The present invention is a flow chart of coding of a group error correction code accompanied by a pseudo noise sequence;

[0030] Figure 3 It is a recovery flow chart of the present invention when data is read out;

[0031] Figure 4 It is a diagram of the construction process of two ~33kb and twelve 6-8kb plasmids in specific embodiment 1 of the present invention;

[0032] Figure 5 The length distribution diagram of 4000 sequencing reads generated by sequencing the 33,559 bp plasmid in specific embodiment 1 of the present invention;

[0033] Figure 6 The length distribution diagram of 4000 sequencing reads generated by sequencing six 6-8 kb plasmids in specific embodiment 1 of the present invention;

[0034] Figure 7 This is a diagram for analyzing the specific situation of data recovery in specific embodiment 1 of the present invention;

[0035] Figure 8 is a codeword sequence error rate distribution diagram in specific embodiment 1 of the present invention;

[0036] Fig. 9 This is a diagram showing the specific results of the parallel experiments in specific embodiment 1 of the present invention;

[0037] Fig.10 It is a diagram of the construction process of two ~43kb and 16 6-8kb plasmids in specific embodiment 1 of the present invention;

[0038] Fig.11 The length distribution diagram of 4000 sequencing reads generated by sequencing 6 to 8 bp plasmids in specific embodiment 2 of the present invention;

[0039] Fig.12 The length distribution diagram of 4000 sequencing reads generated by sequencing a 43,343 bp plasmid in specific embodiment 2 of the present invention;

[0040] Fig.13 This is a diagram for analyzing the specific situation of data recovery in specific embodiment 2 of the present invention;

[0041] Fig.14 is a codeword sequence error rate distribution diagram in specific embodiment 2 of the present invention;

[0042] Fig.15 This is a diagram showing the specific results of the parallel experiments in specific embodiment 2 of the present invention. DETAILED DESCRIPTION

[0043] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention are described in further detail below.

[0044] The present invention is oriented to the field of DNA data storage, and proposes a medium-length DNA design and readout method with pseudo-noise sequence accompanying coding. The basic idea is to encode the information sequence with a group error correction code, and then generate a data DNA sequence with a pseudo-noise sequence; synthesize and assemble the data DNA sequence into medium-length DNA fragments, and perform efficient library construction and third-generation sequencing on them; when reading the data, the sequencing read segments matching the size of the DNA fragments are used, and the pseudo-noise sequence alignment and efficient error correction and erasure decoding processes are combined to achieve fast and error-free data recovery under low sequencing coverage. The medium-length DNA design and readout method with pseudo-noise sequence accompanying coding in the present invention are described in detail below in conjunction with the accompanying drawings.

[0045] Attached Figure 1 The system block diagram of the solution of the present invention is shown, and the solution of the present invention includes the following steps:

[0046] (1) Encode the information sequence using a block error correction code to generate a codeword sequence and a pseudo noise sequence with the same length as the codeword sequence, layer the codeword sequence and the pseudo noise sequence together to construct a two-layer coding sequence, and map the two-layer coding sequence into a data DNA sequence according to a specific mapping rule;

[0047] (2) Synthesize and assemble the data DNA sequences into medium-length DNA fragments of different forms as storage media, perform efficient library construction and third-generation sequencing on the DNA fragments, and generate sequencing reads;

[0048] (3) Preprocess the sequencing reads and demap the processed sequencing reads into damaged codeword sequences and damaged pseudo-noise sequences. The damaged pseudo-noise sequences are compared with known pseudo-noise sequences to determine the alignment positions of the reads and identify insertion / deletion errors therein. The damaged codeword sequences are corrected according to the insertion / deletion error positions. The corrected codeword sequences are judged by majority voting bit by bit to obtain a consistent sequence, which is then sent to a decoder for decoding to restore the original data.

[0049] Among them, as attached Figure 2 As shown, in step (1) of the present invention, the information sequence is encoded using a group error correction code to generate a code word sequence, and a pseudo noise sequence with the same length as the code word sequence is generated, and the code word sequence and the pseudo noise sequence are layered to construct a two-layer coding sequence, and the specific steps of mapping the two-layer coding sequence into a data DNA sequence according to a specific mapping rule are as follows:

[0050] (1.1) A block error correction code (n, k) is used to encode an information sequence of length k to generate a codeword sequence. The length of the encoded codeword sequence is n bits. The block error correction code can be a low-density parity check code (LDPC code) in a communication system, a multi-ary code defined on a high-order Galois field GF(q) (q>2), or an error correction code with a large girth, etc.;

[0051] (1.2) Two m-sequences of different orders are periodically extended respectively, a sequence portion of length n bits is extracted, and a pseudo noise sequence of length n bits is generated by performing bit-by-bit XOR on the sequence portion;

[0052] (1.3) The codeword sequence and the generated pseudo noise sequence are layered together to construct a two-layer coding sequence. One bit is selected from each layer of the coding sequence to form a bit pair.

[0053] (1.4) Map the bit pairs to bases according to the mapping rules: 00→A, 01→T, 10→G, 11→C, and obtain a data DNA sequence of length n.

[0054] Among them, step (2) of the present invention synthesizes and assembles the data DNA sequence into medium-length DNA fragments in different forms as storage media, performs efficient library construction and third-generation sequencing on the DNA fragments, and generates sequencing reads. Specifically, the data DNA sequence is synthesized and assembled into medium-length DNA fragments in various forms such as circular plasmids or artificial chromosomes by means of in vivo assembly and other methods as storage media. The present invention has high robustness to various types of base errors, and can use third-generation sequencing platforms with high error rates (such as Nanopore and PacBio, etc.) for sequencing to generate sequencing reads.

[0055] Among them, as attached Figure 3As shown, in step (3) of the present invention, the sequencing reads are preprocessed, the sequencing reads are demapped into damaged code word sequences and damaged pseudo noise sequences, the damaged pseudo noise sequences are compared with known pseudo noise sequences, the alignment positions of the reads are determined and insertion / deletion errors therein are identified, the damaged code word sequences are corrected according to the insertion / deletion error positions, the corrected code word sequences are subjected to majority voting judgment bit by bit to obtain a consensus sequence, and the consensus sequence is sent to a decoder for decoding, and the specific steps of restoring the original data are as follows:

[0056] (2.1) filtering sequencing reads with large differences in size from DNA fragments;

[0057] (2.2) demapping each base in the retained sequencing read into a bit pair according to the rule of A→00, T→01, G→10, C→11 to obtain an upper-layer damaged codeword sequence and a lower-layer damaged pseudo-noise sequence;

[0058] (2.3) aligning the damaged pseudo-noise sequence with the known pseudo-noise sequence to determine the alignment position of the sequencing reads and identify the insertion / deletion errors therein;

[0059] (2.4) correcting the damaged codeword sequence according to the insertion / deletion error position in the damaged pseudo noise sequence, marking the deletion error position as an erasure error and representing it with a symbol representing an erasure error, and deleting the bit at the insertion error position;

[0060] (2.5) The corrected codeword sequence is subjected to majority voting bit by bit to obtain a consistent sequence, which is then sent to the decoder for decoding to restore the original data.

[0061] Among them, the damaged pseudo-noise sequence is compared with the known pseudo-noise sequence in step (2.3) of the present invention to determine the alignment position of the sequencing reads and identify the insertion / deletion errors therein. Specifically, a dynamic programming algorithm based on global alignment or local alignment, or an index-based sequence alignment algorithm can be used.

[0062] Among them, the step (2.5) of the present invention is to perform majority voting on the modified code word sequence bit by bit to obtain a consistent sequence, and send it to the decoder for decoding to restore the original data. The specific steps are:

[0063] (3.1) For multiple corrected codeword sequences, count the number of bits "1" or "0" in each position, select the bit with the largest number of {0, 1} as the bit at that position in the consistency sequence, and if there is no bit at that position or a majority vote cannot be made, mark it as erased and represent it with a bit indicating an erasure error, to obtain a consistency sequence of length n;

[0064] (3.2) The consistency sequence is sent to the decoder for substitution / erasure error correction to restore the original data. Specific embodiments

[0066] The invention designs a DNA storage method with pseudo-noise sequence encoding, which can realize fast and error-free data recovery at low sequencing coverage. Two specific embodiments are given below to illustrate the feasibility of the invention. Combined with the given embodiments, the purpose, features and advantages of the invention can be further understood. Specific embodiment 1

[0068] The first specific embodiment of the present invention is given below. In this embodiment, two ~33kb long plasmids and 12 6-8kb short plasmids are constructed as DNA storage media. Two 945-byte texts are stored in two plasmids of different length patterns using a highly reliable non-binary LDPC code and a pseudo-noise sequence encoding method. The selected LDPC code is defined in the finite field GF(2 6 ) on LDPC code (22680, 7560), the length of the information bit part is 7,560 bits, the codeword length is 22,680 bits, and the code rate is 1 / 3.

[0069] The specific encoding method of LDPC code and pseudo noise sequence is:

[0070] (1.1) Encode the information sequence of 7,560 bits by LDPC code. The information bit length is 7,560 bits, and the length of the encoded codeword sequence is 22,680 bits.

[0071] (1.2) Extend the periods of two m-sequences of different orders respectively, extract a sequence portion of 22,680 bits in length, and perform bit-by-bit XOR on the sequence to generate a pseudo-noise sequence of 22,680 bits in length;

[0072] (1.3) The encoded codeword sequence is accompanied by a pseudo noise sequence to construct a two-layer coding sequence. One bit is selected from each layer of the coding sequence in turn to form a bit pair.

[0073] (1.4) The bit pairs are mapped to bases according to the mapping rules of 00→A, 01→T, 10→G, 11→C, and a data DNA sequence with a length of 22,680 bp is obtained.

[0074] The data DNA sequence was combined with the vector sequence by in vivo assembly method to construct two ~33 kb long plasmids and 12 6–8 kb short plasmids, such as Figure 4 As shown, efficient library construction and nanopore sequencing were performed to generate third-generation sequencing reads.

[0075] The specific recovery method when reading data is as follows:

[0076] (2.1) For the sequencing reads generated by sequencing two ~33kb plasmids, select the sequencing reads of length 30–35kb. For the sequencing reads generated by sequencing 12 6–8kb plasmids, select the sequencing reads of length 6–8kb;

[0077] (2.2) Demap each base in the sequencing read segment into a bit pair according to the rule of A→00, T→01, G→10, C→11, and obtain a damaged codeword sequence in the upper layer and a damaged pseudo-noise sequence in the lower layer;

[0078] (2.3) Use the hash table index-based sequence alignment software Minimap2 to align the damaged pseudo-noise sequence with the known pseudo-noise sequence, determine the alignment position of the sequencing read segment and identify the insertion / deletion error position therein;

[0079] (2.4) correcting the damaged codeword sequence according to the insertion / deletion error positions in the damaged pseudo noise sequence, that is, converting the bits at the deletion error positions into erasures and representing them with 'E', and deleting the bits at the insertion error positions;

[0080] (2.5) The corrected codeword sequence is judged by majority voting bit by bit to obtain a consistent sequence of 22,680 bp in length, which is then sent to the decoder for decoding to restore the original data.

[0081] Taking a long plasmid of 33,559 bp and six short plasmids of 6–8 kb as examples, two sets of sequencing reads were obtained after efficient library construction and nanopore sequencing. Figure 5 and Figure 6 The length distribution of the 4000 sequencing reads generated is given respectively. Figure 5 It can be seen that there are 1369 reads close to the length of the plasmid (30–35 kb), and their base numbers account for 61.12% of the total base numbers. Figure 6 It can be seen that using the same library construction method for short plasmids, most of the reads are also located in a narrow length range (6–8 kb) that matches the plasmid size.

[0082] For long noisy reads close to the length of the plasmid, this example conducted multiple independent experiments at different sequencing coverages. In order to verify the reliability of data recovery and determine the minimum sequencing coverage required for error-free data recovery, we selected a coverage of 1–6× in actual testing. Figure 7The data recovery of 33kb and 6-8kb plasmids at different sequencing coverages is given. It can be seen that the plasmids of two different length modes can achieve error-free data recovery at a minimum of 3× sequencing coverage, and some independent experiments can still be successfully recovered at 1× sequencing coverage, which verifies the high reliability of the encoding method of LDPC code and pseudo-noise sequence proposed in the scheme of the present invention, and can achieve fast readout and data recovery in the case of close to single-copy single molecule. Figure 8 The error rates of the codeword sequences after insertion / deletion errors are corrected for a 33kb plasmid at different coverages are given. It can be seen that at 3× sequencing coverage, the erasure rate of the codeword sequence after insertion / deletion error correction is 1.9% and the error rate is 2%, which is within the error correction capability of non-binary LDPC codes. To demonstrate the consistency of the proposed scheme, parallel experiments were performed using different data sets, and the results were similar. The specific results are shown in the figure. Fig. 9 shown. Specific embodiment 2

[0084] The second specific embodiment of the present invention is given below. In this embodiment, two ~43kb long plasmids and 16 6-8kb short plasmids are constructed. A 4,032-byte text is stored in two plasmids of different length patterns using a highly reliable non-binary LDPC code and a pseudo-noise sequence encoding method. The selected LDPC code is defined in the finite field GF(2 6 ) on LDPC code (64512, 32256), the length of the information bit part is 32,256 bits, the codeword length is 64,512 bits, and the code rate is 1 / 2.

[0085] The encoding method of LDPC code and pseudo noise sequence is as follows:

[0086] (1.1) Encode the information sequence of 32,256 bits by LDPC code. The information bit length is 32,256 bits, and the length of the encoded codeword sequence is 64,512 bits.

[0087] (1.2) Extend the periods of two m-sequences of different orders respectively, extract the sequence part with a length of 64,512 bits, and perform bit-by-bit XOR to generate a pseudo-noise sequence with a length of 64,512 bits;

[0088] (1.3) The encoded codeword sequence is layered with a pseudo noise sequence of 64,512 bits to construct a two-layer coding sequence. One bit is selected from each layer of the coding sequence to form a bit pair.

[0089] (1.4) Map the bit pairs to bases according to the mapping rules of 00→A, 01→T, 10→G, 11→C, and obtain a data DNA sequence with a length of 64,512 bp.

[0090] The data DNA sequence was combined with the vector sequence by in vivo assembly method to construct two ~43 kb plasmids and 16 6–8 kb plasmids, such as Fig.10 As shown, efficient library construction and nanopore sequencing were performed to generate third-generation sequencing reads.

[0091] The specific recovery method when reading data is as follows:

[0092] (2.1) For the sequencing reads generated by sequencing the two ~43 kb plasmids, the sequencing reads of length 40–45 kb were selected, and for the sequencing reads generated by sequencing the 16 6–8 kb plasmids, the sequencing reads of length 6–8 kb were selected;

[0093] (2.2) Demap each base in the sequencing read segment into a bit pair according to the rule of A→00, T→01, G→10, C→11, and obtain a damaged codeword sequence in the upper layer and a damaged pseudo-noise sequence in the lower layer;

[0094] (2.3) Use the hash table index-based sequence alignment software Minimap2 to align the damaged pseudo-noise sequence with the known pseudo-noise sequence to determine the location of the sequencing reads and identify the insertion / deletion error locations;

[0095] (2.4) correcting the damaged codeword sequence according to the insertion / deletion error positions in the damaged pseudo noise sequence, that is, converting the bits at the deletion error positions into erasures and representing them with 'E', and deleting the bits at the insertion error positions;

[0096] (2.5) The corrected codeword sequence is judged by majority voting bit by bit to obtain a consistent sequence of 64,512 bp in length, which is then sent to the decoder for decoding to restore the original data.

[0097] Two ~43 kb plasmids and 16 6–8 kb plasmids were efficiently constructed and sequenced by nanopore sequencing, resulting in a total of 4 sets of sequencing reads. Fig.11 and Fig.12 Represents the length distribution of 4000 sequencing reads generated by sequencing 6 ~8kb and 1 ~43kb plasmids, respectively. It can be seen that after efficient library construction of plasmids with different length patterns, most of the read lengths are within a narrow length range that matches the plasmid size. Reads close to the plasmid size are selected for data recovery. This example performs multiple repeated experiments at different sequencing coverages to determine the minimum sequencing coverage required for data recovery and evaluate the error correction performance of LDPC codes with different code rates. In the actual test, we selected a sequencing coverage of 1–6×. Fig.13 The data recovery conditions at different sequencing coverages are given. It can be seen that error-free data recovery can be achieved at the lowest 3× sequencing coverage, which further verifies the effectiveness of the encoding method adopted in the present invention. Fig.14 The error rates of the codeword sequences of 16 short plasmids after insertion / deletion error correction at different sequencing coverages are given. It can be seen that at 3× sequencing coverage, the erasure rate of the codeword sequence is 2.8% and the error rate is 1.7%, which is within the error correction range of non-binary LDPC codes with a code rate of 1 / 2, verifying the efficient error correction performance of the LDPC code used. Parallel experiments were conducted using different data sets, and the results are shown in Figure 2. Fig.15 It can be seen that using different data sets, two plasmids of different lengths can achieve error-free recovery at 3× sequencing coverage, proving that the scheme of the present invention can achieve reliable data storage at extremely low sequencing coverage.

[0098] Those skilled in the art will appreciate that the accompanying drawing is only a schematic diagram of a preferred embodiment, and the serial numbers of the embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0099] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for storing pseudo-noise sequence-encoded DNA, characterized in that: The method comprises the following steps: (1) Using a group error correction code to group encode the data to be stored by the user, generate a code word bit sequence, and generate a binary pseudo noise sequence with the same length as the code word bit sequence, divide the error correction coded code word sequence and the pseudo noise sequence into two layers, and construct two layers of coding sequences with exactly the same length, and map the two layers of coding sequences into data DNA sequences according to a specific bit pair to base mapping rule; (2) Connecting the data DNA sequence with the vector sequence, using chemical synthesis or enzymatic synthesis, assembling into different forms of DNA fragment entities as storage media, building libraries and performing third-generation sequencing on the DNA fragments to generate sequencing reads; (3) Demapping the sequencing reads into damaged codeword sequences and damaged pseudo-noise sequences, determining the alignment position of the reads and identifying the insertion / deletion errors therein by aligning the damaged pseudo-noise sequences with the known pseudo-noise sequences, correcting the damaged codeword sequences according to the insertion / deletion error positions, and performing a majority vote on the corrected codeword sequences bit by bit to obtain a consensus sequence, which is then sent to a decoder for decoding to restore the original data.

2. The method for storing pseudo-noise sequence-combined coded DNA according to claim 1, characterized in that: The method uses a group error correction code to group encode the data to be stored by the user, generates a code word bit sequence, and generates a binary pseudo noise sequence with the same length as the code word bit sequence, divides the error correction coded code word sequence and the pseudo noise sequence into two layers of bit-by-bit correspondence, constructs two layers of coding sequences with exactly the same length, and maps the two layers of coding sequences into data DNA sequences according to a specific bit pair to base mapping rule. The specific steps are as follows: (1.1) A block error correction code (n, k) is used to encode an information sequence of length k to generate an error correction coded codeword sequence. The length of the coded codeword sequence is n bits. The block error correction code uses a low-density parity check code with parameters (n, k), a multi-level low-density parity check code defined on a high-order Galois field GF(q), q>2, a low-density parity check code with a large girth, and a Reed-Solomon code, where k is the length of the information sequence, i.e., the number of information bits, and n is the length of the coded error correction codeword; (1.2) The pseudo-random sequence accompanying the coded codeword bit by bit is obtained by periodically extending two m sequences of different orders to obtain two sequences of length n bits, and then performing bit-by-bit XOR on the sequences intercepted by the two periodic extensions of length n bits to obtain a pseudo-noise sequence of length n bits; (1.3) constructing two-layer coding sequences from the error correction coding codeword sequence and the generated pseudo noise sequence of length n bits, taking out one bit from each layer one by one, and selecting one bit from each layer of coding sequence one by one to form a bit pair; (1.4) Map the bit pairs to bases according to the mapping rules: 00→A, 01→T, 10→G, 11→C, and obtain a data DNA sequence of length n.

3. The method for storing pseudo-noise sequence-combined coded DNA according to claim 1, characterized in that: The data DNA sequence is connected to the vector sequence, and different forms of DNA fragment entities are assembled by chemical synthesis or enzymatic synthesis. As the storage medium, the DNA fragments are library-built and sequenced for the third generation to generate sequencing reads. The specific description is: The data DNA sequence is connected to the vector sequence, and then synthesized and assembled into medium-length DNA fragments of different forms as storage media, and the DNA fragments are subjected to library construction and third-generation sequencing to generate sequencing reads. Specifically, the data DNA sequence is synthesized and assembled into medium-length DNA fragments of circular plasmids or artificial chromosomes with the help of an in vivo assembly method, and used as storage media. Sequencing is performed using a third-generation sequencing platform to generate sequencing reads.

4. The method for storing pseudo-noise sequence-combined coded DNA according to claim 1, characterized in that: The sequencing reads are demapped into damaged codeword sequences and damaged pseudo-noise sequences, the damaged pseudo-noise sequences are compared with known pseudo-noise sequences, the alignment positions of the reads are determined and insertion / deletion errors therein are identified, the damaged codeword sequences are corrected according to the insertion / deletion error positions, the corrected codeword sequences are subjected to majority voting bit by bit to obtain a consensus sequence, and the consensus sequence is sent to a decoder for decoding to restore the original data. The specific steps are as follows: (4.1) Demap each base in the sequencing read into a bit pair according to the rule of A→00, T→01, G→10, C→11, where the upper bit sequence is the damaged codeword sequence and the lower bit sequence is the damaged pseudo noise sequence; (4.2) The known pseudo-noise sequence of length n at the read end is extended to twice the length, and the damaged pseudo-noise sequence is compared with the extended twice-length sequence to obtain the optimal matching position of the damaged pseudo-noise sequence, obtain the alignment position of the sequencing read segment and identify the specific position of the insertion / deletion error of the damaged pseudo-noise sequence; (4.3) Based on the fact that the positions where insertion and deletion errors occur in the damaged pseudo-noise sequence are exactly the same as the positions where insertion and deletion errors occur in the coded codeword, the damaged codeword sequence is corrected according to the positions of insertion / deletion errors in the damaged pseudo-noise sequence, that is, a mark is added to the position where the deletion error occurs, marked as an erasure error, represented by a symbol representing an erasure error, and the bit at the position where the insertion error occurs is directly removed, thereby correcting the insertion error and converting the deletion error into an erasure; (4.4) The corrected codeword sequence is placed in the correct position obtained by comparison. Considering that some positions are repeatedly overlapped by multiple codeword sequences, a majority decision is made on all the corrected codeword sequences at each position bit by bit to obtain a consistent sequence. The uncovered codeword parts are marked as erased, thereby obtaining a complete block-coded codeword sequence, which is then sent to the decoder for decoding to restore the original data.

5. The method for storing pseudo-noise sequence-combined coded DNA according to claim 4, characterized in that: The modified codeword sequence is placed at the correct position obtained by comparison. Considering that some positions have multiple codeword sequences that are repeatedly covered, a majority decision is made on all the modified codeword sequences at each position bit by bit to obtain a consistent sequence. The uncovered codeword parts are marked as erased, thereby obtaining a complete block-coded codeword sequence, and sending it to a decoder for decoding to restore the original data. The specific steps are as follows: (5.1) For multiple corrected codeword sequences, count the number of bits "1" or "0" in each position, select the value with the largest number of {0, 1} as the bit value at that position in the consistency sequence, and if there is no bit at that position or a majority vote cannot be made, mark it as an erasure error and represent it with a bit symbol representing an erasure error, and finally obtain a consistency sequence of length n; (5.2) The consistency sequence is sent to the decoder of the error correction code for error correction including substitution and erasure to restore the original data. The decoder of the error correction code is matched with the encoding method, that is, the decoder of the low-density parity-check code with parameters (n, k), the multi-base low-density parity-check code defined on the high-order Galois field GF(q), q>2, the low-density parity-check code with large girth, and the Reed-Solomon code.