Lesion-specific data screening and processing method, device, equipment and storage medium

CN114093424BActive Publication Date: 2024-11-19BAMBOO STONE BIOTECHNOLOGY (SUZHOU) CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111439901.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-11-30
Publication Date
2024-11-19
Estimated Expiration
2041-11-30

AI Technical Summary

Technical Problem

The sequencing data of DNA in lesion tissues is mixed with a large number of non-lesion cells, resulting in a decrease in neural network training quality and a decrease in recognition accuracy.

Method used

By obtaining lesion sample data and control sample data, differential methylation regions are found, methylation continuity scores are calculated, lesion-specific data are screened, and annotated to train the neural network.

Benefits of technology

This improves the training quality of neural networks and the final recognition accuracy, reducing the interference of non-lesion cell data on lesion cell recognition.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114093424B_ABST
    Figure CN114093424B_ABST
Patent Text Reader

Abstract

The present invention provides a lesion-specific data screening and processing method, device, equipment and storage medium, the method comprising: pre-processing lesion sample data and control sample data to obtain a first lesion sample read data set and a first control sample read data set; based on the first lesion sample read data set and the first control sample read data set, finding the target differential methylation region between the two sets, and obtaining a second lesion sample read data set and a second control sample read data set representing each target differential methylation region; calculating the methylation continuity score of each second lesion sample read data and the second control sample read data set, and using the second control sample read data set as a reference and based on the methylation continuity score, lesion-specific data is screened out from the second lesion sample read data set. The present invention is conducive to improving the training quality and recognition accuracy of the neural network.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of bioinformatics, and in particular to a method, device, equipment and storage medium for screening and processing lesion-specific data. Background Art

[0002] The reads obtained by high-throughput DNA methylation sequencing of diseased tissues obtained in clinical work (such as surgical resection) are often mixed with a considerable proportion of data from non-lesional cells, which brings noise to downstream data analysis. For example, DNA from diseased tissues and normal cells are subjected to high-throughput DNA methylation sequencing respectively, and the data from both are labeled to train the neural network. Ultimately, it is expected that the trained neural network can automatically identify the sequencing data from plasma circulating DNA (cfDNA) and distinguish whether the new read data is from diseased cells or normal cells.

[0003] However, during this training process, the sequencing data from the diseased tissue DNA contains not only data from diseased cells, but also a large amount of data from normal cells. This will bring noise and confusion to the training of the neural network, reducing the training quality of the neural network and the final recognition accuracy. Summary of the invention

[0004] The present invention provides a lesion-specific data screening and processing method, device, equipment and storage medium, which are used to solve the problem in the prior art that the sequencing data of lesion tissue DNA will bring noise and confusion to the training of neural network.

[0005] In a first aspect, an embodiment of the present invention provides a method for screening and processing lesion-specific data, characterized in that the method comprises:

[0006] Acquire lesion sample data and control sample data, and preprocess the lesion sample data and the control sample data to obtain a first lesion sample read data set and a first control sample read data set, wherein each first lesion sample read data and each first control sample read data represent the base arrangement order on a DNA sequence and contain methylation state information of each cytosine on the DNA sequence;

[0007] Based on the first lesion sample read data set and the first control sample read data set, searching for at least one target differential methylation region between the two sets, and obtaining a second lesion sample read data set and a second control sample read data set representing each target differential methylation region;

[0008] The methylation continuity score of each second lesion sample read data and the second control sample read data is calculated, and the lesion-specific data is screened out from the second lesion sample read data set with the second control sample read data set as a reference and according to the methylation continuity score.

[0009] In one embodiment of the present invention, the method further comprises:

[0010] After the selected lesion-specific data are labeled, a preset neural network is trained so that the trained neural network can predict the probability of lesion occurrence based on the input samples.

[0011] In one embodiment of the present invention, the preprocessing of the lesion sample data and the control sample data includes:

[0012] Dividing the lesion sample data and the control sample data into a training set and a test set, wherein the sample data in the training set is used to screen out the lesion-specific data and to train a neural network, and the sample data in the test set is used to test the trained neural network;

[0013] Among them, the lesion sample data includes DNA data of lesion tissues from a preset number of patients, and the control sample data includes DNA data of non-lesion tissues next to the lesion tissues of the preset number of patients, and cfDNA data of a preset number of healthy people and high-risk people.

[0014] In one embodiment of the present invention, the step of acquiring the lesion sample data and the control sample data, and preprocessing the lesion sample data and the control sample data to obtain a first lesion sample read data set and a first control sample read data set further includes:

[0015] The DNA molecules of each sample are broken into DNA fragments of a preset length, and the DNA fragments are respectively converted by enzymatic or salt methods to construct a DNA library;

[0016] Whole genome methylation sequencing is performed on the DNA library to obtain the first lesion sample read data set and the first control sample read data set.

[0017] In one embodiment of the present invention, performing whole genome methylation sequencing on the DNA library to obtain the first lesion sample read data set and the first control sample read data set comprises:

[0018] After whole-genome methylation sequencing of the DNA library, data filtering is performed on the sequencing results based on preset rules;

[0019] The filtered data is compared with a preset reference genome and the methylation status is identified to obtain the first diseased sample read data set and the first control sample read data set.

[0020] In one embodiment of the present invention, the step of searching for at least one target differentially methylated region between the first lesion sample read data set and the first control sample read data set based on the two sets, and obtaining a second lesion sample read data set and a second control sample read data set representing each target differentially methylated region comprises:

[0021] Searching for differentially methylated regions between first lesion sample read data and first control sample read data corresponding to the lesion tissues and non-lesion tissues of the preset number of patients, respectively, to obtain a first set;

[0022] Finding differentially methylated regions between the first lesion sample read data and the first control sample read data corresponding to the lesion tissues of the preset number of patients and the DNA data of the preset number of high-risk people, respectively, to obtain a second set;

[0023] Taking the intersection of the first set and the second set as the target differentially methylated region;

[0024] According to the target differentially methylated regions, a second lesion sample read data set and a second control sample read data set representing each target differentially methylated region are obtained.

[0025] In one embodiment of the present invention, the step of calculating the methylation continuity score of each second lesion sample read data and the second control sample read data, and using the second control sample read data set as a reference and screening out lesion-specific data from the second lesion sample read data set according to the methylation continuity score comprises:

[0026] Calculate the methylation continuity score of each second lesion sample read data and the methylation continuity score of each read data corresponding to the healthy population in the second control sample read data set, and mark the maximum and minimum values ​​of the methylation state continuity scores of all second control sample read data corresponding to the healthy population as Smax and Smin, respectively;

[0027] If the target differentially methylated region is a low-methylated region, removing the read data with a methylation state continuity score greater than or equal to Smin in the second lesion sample read data set;

[0028] If the target differentially methylated region is a high methylated region, the read data with a methylation state continuity score less than or equal to Smax in the second lesion sample read data set is removed.

[0029] In one embodiment of the present invention, the step of calculating the methylation continuity score of each second lesion sample read data and the second control sample read data, and using the second control sample read data set as a reference and filtering out lesion-specific data from the second lesion sample read data set according to the methylation continuity score further includes:

[0030] Calculating the methylation continuity score of each second lesion sample read data and the methylation continuity score of each read data corresponding to the high-risk population in the second control sample read data set, and sorting the methylation status continuity scores of all second control sample read data corresponding to the high-risk population from high to low;

[0031] If the target differentially methylated region is a low-methylated region, a preset percentage of read data after ranking the methylation state continuity score in the second control sample read data set corresponding to the high-risk population is removed, and the minimum value of the methylation state continuity score of the remaining read data is recorded as Smin, and the read data with a methylation state continuity score greater than or equal to Smin in the second lesion sample read data set is removed;

[0032] If the target differentially methylated region is a high methylated region, the read data with a preset percentage of methylation status continuity scores ranked in the top of the second control sample read data set corresponding to the high-risk population are removed, and the maximum value of the methylation status continuity scores of the remaining read data is recorded as Smax, and the read data with a methylation status continuity score less than or equal to Smax in the second lesion sample read data set are removed.

[0033] In one embodiment of the present invention, the step of taking the second control sample read data set as a reference and screening out lesion-specific data from the second lesion sample read data set according to the methylation continuity score further comprises:

[0034] After the removal operation, for each target differentially methylated region, according to a preset indicator function and the read data in all the second lesion sample read data sets corresponding to the target differentially methylated region, calculating an indicator value associated with the target differentially methylated region;

[0035] All index values ​​are sorted from large to small, and the target differential methylation regions associated with the preset index values ​​with the highest sorting are selected as the targeted differential methylation regions;

[0036] The read data in the second lesion sample read data set after the removal operation corresponding to each targeted differentially methylated region is determined as the lesion-specific data.

[0037] In a second aspect, an embodiment of the present invention further provides a lesion-specific data screening and processing device, the device comprising:

[0038] A preprocessing module, used to obtain lesion sample data and control sample data and preprocess the lesion sample data and the control sample data to obtain a first lesion sample read segment data set and a first control sample read segment data set, wherein each first lesion sample read segment data and each first control sample read segment data represent the base arrangement order on a DNA sequence and contain methylation status information of each cytosine on the DNA sequence;

[0039] A search module, configured to search for at least one target differential methylation region between the first lesion sample read data set and the first control sample read data set based on the first lesion sample read data set and the first control sample read data set, and obtain a second lesion sample read data set and a second control sample read data set representing each target differential methylation region;

[0040] A screening module is used to calculate the methylation continuity score of each second lesion sample read data and the second control sample read data, and to screen out lesion-specific data from the second lesion sample read data set based on the methylation continuity score and the second control sample read data set as a reference.

[0041] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the lesion-specific data screening and processing method as described in any one of the first aspects above are implemented.

[0042] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the lesion-specific data screening and processing method as described in any one of the first aspects above.

[0043] The present invention provides a lesion-specific data screening and processing method, device, equipment and storage medium. By searching for differential methylation regions between a first lesion sample read data set and a first control sample read data set, a second lesion sample read data set and a second control sample read data set representing each target differential methylation region are obtained. The lesion-specific data is screened out from the second lesion sample read data set using the second control sample read data set as a reference and according to the methylation continuity score. The lesion-specific data is then annotated and then the neural network is trained, which is beneficial to improving the training quality of the neural network and the final recognition accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other embodiments can be obtained based on these drawings without paying creative work.

[0045] Figure 1 is a flow chart of a method for screening and processing lesion-specific data provided by an embodiment of the present invention;

[0046] Figure 2 is a schematic diagram of the methylation status continuity score of each read segment data provided by an embodiment of the present invention;

[0047] FIG3( a) is one of the schematic diagrams of low methylation regions and high methylation regions provided in an embodiment of the present invention;

[0048] FIG3( b ) is a second schematic diagram of a low methylation region and a high methylation region provided in an embodiment of the present invention;

[0049] Figure 4 is a schematic diagram of the structure of a lesion-specific data screening and processing device provided by an embodiment of the present invention;

[0050] Figure 5 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein.

[0053] The technical background of the present invention is described below:

[0054] The lesion tissue obtained by clinical work (such as surgical resection) contains not only lesion cells, but also often mixed with non-lesion cells. The non-lesion cells mixed in the lesion tissue mainly include non-lesion tissue cells, immune inflammatory cells, matrix cells, vascular endothelial cells, etc. The proportion of lesion cells in the lesion tissue is the lesion purity. For example, the lesion purity of the lesion tissue specimen obtained by standard surgical operation is usually less than 70%.

[0055] Although there are some algorithms for calculating lesion purity, these algorithms can only estimate the overall possible lesion purity based on model assumptions. There is no efficient denoising method for screening lesion cell-specific data from DNA methylation sequencing data of lesion tissues.

[0056] In order to solve the problem in the prior art that the sequencing data of diseased tissue DNA will bring noise and confusion to the training of neural networks, the present invention provides a lesion-specific data screening and processing method, device, equipment and storage medium, by obtaining a first lesion sample read data set and a first control sample read data set, finding the differential methylation region between the two sets, and obtaining a second lesion sample read data set and a second control sample read data set representing each target differential methylation region, and using the second control sample read data set as a reference and according to the methylation continuity score, lesion-specific data is screened out from the second lesion sample read data set, and then the lesion-specific data is annotated and then the neural network is trained, which is beneficial to improving the training quality of the neural network and the final recognition accuracy.

[0057] Combine the following Figure 1-Figure 5 The present invention describes a method, device, equipment and storage medium for screening and processing lesion-specific data provided by an embodiment of the present invention.

[0058] Figure 1 FIG. 1 is a flow chart of a method for screening and processing lesion-specific data provided by an embodiment of the present invention. Figure 1 An embodiment of the present invention provides a method for screening and processing lesion-specific data, the method comprising:

[0059] Step 101, obtaining lesion sample data and control sample data and preprocessing the lesion sample data and the control sample data to obtain a first lesion sample read data set and a first control sample read data set, wherein each first lesion sample read data and each first control sample read data represent the base arrangement order on a DNA sequence and contain the methylation status information of each cytosine on the DNA sequence.

[0060] Step 102: searching for at least one target differential methylation region between the first lesion sample read data set and the first control sample read data set based on the two sets, and obtaining a second lesion sample read data set and a second control sample read data set representing each target differential methylation region.

[0061] Step 103 , calculating the methylation continuity score of each second lesion sample read data and the second control sample read data, and filtering out lesion-specific data from the second lesion sample read data set using the second control sample read data set as a reference and according to the methylation continuity score.

[0062] Exemplarily, after step 103, the method further includes:

[0063] Step 104 , after labeling the screened lesion-specific data, a preset neural network is trained so that the trained neural network can predict the probability of lesion occurrence based on the input samples.

[0064] The above steps 101 to 104 are described in detail below.

[0065] In the above step 101, the preprocessing of the lesion sample data and the control sample data includes:

[0066] Step 1011, divide the lesion sample data and the control sample data into a training set and a test set, the sample data in the training set is used to screen out the lesion-specific data and to train the neural network, and the sample data in the test set is used to test the trained neural network.

[0067] The lesion sample data include DNA data of lesion tissues from a preset number of patients, and the control sample data include DNA data of non-lesion tissues adjacent to the lesion tissues of the preset number of patients, and DNA data of a preset number of healthy people and high-risk people.

[0068] Exemplarily, the training set includes paired tissue samples of diseased tissue and non-lesioned tissue from a preset number of cancer patients (e.g., 50 cases of esophageal cancer), as well as blood samples from a preset number of healthy people (e.g., 20 cases) and a preset number of high-risk people (e.g., 50 cases); the test set includes blood samples from a preset number of high-risk people and preset number of cancer patients.

[0069] The above blood samples (including blood samples from healthy people and blood samples from high-risk people) are centrifuged to obtain plasma, and corresponding plasma samples are obtained, that is, plasma samples from healthy people and plasma samples from high-risk people are obtained.

[0070] The above-mentioned tissue samples (including tissue samples of diseased tissues of cancer patients and tissue samples of non-lesioned tissues paired with the diseased tissues) and plasma samples are stored separately, for example, the tissue samples and the plasma samples are stored at -80°C.

[0071] Exemplarily, in the above step 101, the preprocessing of the lesion sample data and the control sample data to obtain a first lesion sample read data set and a first control sample read data set further includes:

[0072] Step 1011, breaking the DNA molecules of each sample into DNA fragments of a preset length, and performing enzymatic or salt conversion on the DNA fragments respectively to construct a DNA library.

[0073] The DNA molecules in each sample are physical DNA molecules. Since the DNA molecule chains extracted from cells or plasma are very long, they need to be broken into short fragments by ultrasonic vibration or other physical and chemical means.

[0074] Among them, cfDNA stands for circulating DNA, also known as free DNA (circulating free DNA or cell free DNA, cfDNA), which refers to short fragments of partially degraded DNA that are released into the human peripheral blood during cell apoptosis or necrosis and are free from cells.

[0075] Among them, ctDNA stands for circulating diseased DNA, which refers to the DNA fragments in cfDNA that come from diseased cells.

[0076] For example, genomic DNA is extracted from tissue samples using the DNeasy Blood and Tissue Kit (Qiagen).

[0077] For example, QIAamp Circulating Nucleic Acid Kit (Qiagen) is used to extract circulating DNA (cfDNA) from plasma samples.

[0078] For example, genomic DNA extracted from tissues is fragmented into fragments of about 200 bp in length using a Bioruptor ultrasonic disruptor.

[0079] For example, the DNA fragments and the circulating DNA are enzymatically converted and the library is constructed using the NEBNext Enzymatic Methyl-seq Kit (NEB).

[0080] For example, bisulfite or enzymes may be used to transform diseased tissue DNA.

[0081] Step 1012: perform whole genome methylation sequencing on the DNA library to obtain the first diseased sample read data set and the first control sample read data set.

[0082] For example, the DNA library is subjected to whole genome methylation sequencing using a second generation high throughput sequencer Novasek 6000. The sequencing strategy is a double-end 150 bp read length. The sequencing depth of each sample is about 10×.

[0083] Exemplarily, fastp software is used to perform quality control and data filtering on the fastq file containing reads data in the sequencing results: (1) filter out reads with a base quality value less than 15 accounting for more than 40%; (2) filter out reads containing more than 5 bases N; (3) remove the adapter sequence; (4) remove the sequence of 15 bp in length at the back end of the first read (read1) and the front end of the second read (read2); (5) filter out reads with a length of less than 36 bp.

[0084] It should be noted that reads refers to the base sequence data on each DNA molecule read by the second-generation high-throughput sequencer, also known as read segments.

[0085] Illustratively, in the above step 1012, performing whole genome methylation sequencing on the DNA library to obtain the first lesion sample read data set and the first control sample read data set includes:

[0086] Step 10121, after performing whole genome methylation sequencing on the DNA library, data filtering is performed on the sequencing results based on preset rules.

[0087] Step 10122, compare the filtered data with a preset reference genome and identify the methylation status to obtain the first diseased sample read data set and the first control sample read data set.

[0088] For example, use Bismark software to align the filtered reads to the human reference genome hg38. Use the deduplication tool in Bismark software to remove PCR (polymerase chain reaction) duplications in the reads based on information such as the alignment position of the reads on the reference genome. Use the methylation extractor tool in Bismark software to obtain the methylation status information of each cytosine base site in the genome. For a cytosine site, the ratio of the number of reads that support methylation of the site to the total number of reads covering the site is the methylation rate of the site.

[0089] Among them, DNA methylation refers to the chemical modification process in which a methyl group is obtained on the 5th carbon atom of cytosine in the nucleotide.

[0090] The CpG site refers to a tightly linked cytosine nucleotide-phosphodiester bond-guanine nucleotide site on a DNA sequence.

[0091] Among them, CpG site methylation refers to the methylation of cytosine in CpG dinucleotides. Methylation of cytosine in DNA usually occurs in CpG sites.

[0092] Exemplarily, the first lesion sample read data and the first control sample read data are high-throughput DNA methylation sequencing data.

[0093] Exemplarily, high-throughput DNA methylation sequencing refers to obtaining the base sequence on the DNA sequence and the methylation status of each cytosine therein by means of second-generation high-throughput sequencing. The data is presented in the form of reads. Before sequencing, the DNA needs to be treated with bisulfite or enzymes in advance, and the unmethylated cytosine in the DNA molecule will be converted into uracil, so that methylated and unmethylated cytosine can be distinguished from the sequencing results, thereby obtaining the methylation status of each cytosine. Including but not limited to whole genome methylation sequencing and high-throughput targeted methylation sequencing.

[0094] The above step 102 is described in detail below.

[0095] In the above step 102, the step of searching for at least one target differentially methylated region between the first lesion sample read data set and the first control sample read data set based on the two sets, and obtaining a second lesion sample read data set and a second control sample read data set representing each target differentially methylated region includes:

[0096] Step 1021 , searching for differentially methylated regions between first lesion sample read data and first control sample read data corresponding to the lesion tissues and non-lesion tissues of the preset number of patients, respectively, to obtain a first set.

[0097] For example, the R language software package dmrseq can be used to find differentially methylated regions (DMRs) based on a generalized least squares (GLS) regression model.

[0098] DMR stands for Differentially Methylated Region (DMR for short). It refers to the genomic region where there is a significant difference in methylation rate between two groups of samples from different sources (such as samples from diseased tissue and non-lesioned tissue). The DMR with increased methylation rate in diseased tissue relative to non-lesioned tissue is called a hypermethylated region, and the reverse is called a hypomethylated region.

[0099] For example, tissue samples of diseased tissue and non-lesioned tissue are divided into different groups, and based on the corresponding first diseased sample read data and first control sample read data obtained from different groups, DMRs are searched between different groups using the software package dmrseq, and the found DMRs are called the first set.

[0100] Step 1022, searching for differentially methylated regions between the first lesion sample read data and the first control sample read data corresponding to the lesion tissues of the preset number of patients and the cfDNA data of the preset number of high-risk people, respectively, to obtain a second set.

[0101] Step 1023: taking the intersection of the first set and the second set as the target differentially methylated region.

[0102] Step 1024: According to the target differentially methylated regions, a second lesion sample read data set and a second control sample read data set representing each target differentially methylated region are obtained.

[0103] For example, in order to reduce the potential impact of other variables, two covariates can be set during the analysis, namely covariate A and covariate B. Covariates are input parameters of the dmrseq software package. Age (numeric value) and gender (character code) are used as covariate A and covariate B of the above model, respectively.

[0104] Therefore, high-throughput methylation sequencing is performed on DNA from diseased tissue samples and non-lesion samples (such as paracancerous tissue, plasma, etc.), respectively. The obtained high-throughput methylation sequencing data is used to find differential methylation regions (DMRs) between the DNA of diseased tissue samples and non-lesion samples.

[0105] The above step 103 is described in detail below.

[0106] By calculating the methylation continuity score (MCS) of each read in the DMR region, based on the methylation continuity score, the reads from the diseased tissue DNA are filtered using the reads from the plasma cfDNA of the non-lesioned population to obtain the disease-specific data. This reduces the data noise caused by the DNA from non-lesioned cells mixed in the diseased tissue DNA.

[0107] Among them, lesion-specific data refers to the reads generated by high-throughput DNA methylation sequencing that are unique to diseased cells but not in other normal cells.

[0108] Among them, the calculation formula of the methylation continuity score (MCS) of a read is:

[0109]

[0110] Where L represents the number of CpG sites on the read, n i It means that a continuous block of i methylated CpG sites on a read is defined as a methylation state continuity block, n i Indicates the number of corresponding blocks on read.

[0111] For example, the value range of MCS is 0 to 1. The larger the MCS value, the higher the methylation level on the read, and the methylated CpG sites tend to be distributed continuously on the read, and are less separated by non-methylated CpG sites.

[0112] Figure 2 is a schematic diagram of the methylation status continuity score of each read segment data provided by an embodiment of the present invention, such as Figure 2As shown in the figure. Each horizontal line in the figure represents a read. The solid circles represent methylated CpG sites, and the hollow circles represent unmethylated CpG sites. The numbers on the left (1, 0.44, 0.22, 0) are the methylation status continuity scores (MCS) of each read.

[0113] FIG3(a) is one of the schematic diagrams of the hypomethylated region and the hypermethylated region provided in an embodiment of the present invention, and FIG3(b) is the second schematic diagram of the hypomethylated region and the hypermethylated region provided in an embodiment of the present invention. FIG3(a) and FIG3(b) show methods for screening lesion-specific reads for DMRs in hypomethylated regions and hypermethylated regions, respectively. The diagrams on the left side of FIG3(a) and the left side of FIG3(b) are methods for hypomethylated regions, and the diagrams on the right side of FIG3(a) and the right side of FIG3(b) are methods for hypermethylated regions. The dark diagonal filled portion in the column is the portion to be removed, and the light dot filled portion is the portion to be retained.

[0114] Specifically, in the above step 103, the methylation continuity score of each second lesion sample read data and the second control sample read data is calculated, and the second control sample read data set is used as a reference and the lesion-specific data is screened out from the second lesion sample read data set according to the methylation continuity score, including:

[0115] Step 1031, respectively calculate the methylation continuity score of each second lesion sample read data and the methylation continuity score of each read data corresponding to the healthy population in the second control sample read data set, and mark the maximum and minimum values ​​of the methylation status continuity scores of all second control sample read data corresponding to the healthy population as Smax and Smin, respectively.

[0116] For example, as shown in Figure 3(a), in the left hypomethylated region, all the read data from the patient's diseased tissue DNA (corresponding to the diseased tissue DNA on the horizontal axis), all the read data from the healthy population's cfDNA (corresponding to the healthy population's cfDNA on the horizontal axis), and the vertical axis is the distribution range of the MCS values ​​of all the read data (reads). The minimum value Smin of the methylation status continuity score in all the read data of the healthy population is marked in Figure 3(a).

[0117] For example, as shown in Figure 3(a), in the high methylation region on the right, all the read data from the patient's diseased tissue DNA (corresponding to the diseased tissue DNA on the horizontal axis), all the read data from the healthy population's cfDNA (corresponding to the healthy population's cfDNA on the horizontal axis), and the vertical axis is the distribution range of the MCS values ​​of all the read data (reads). In Figure 3(a), the maximum value Smax of the methylation state continuity score in all the read data of the healthy population is marked.

[0118] Step 1032: If the target differentially methylated region is a low-methylated region, remove the read data with a methylation state continuity score greater than or equal to Smin in the second lesion sample read data set.

[0119] For example, as shown in FIG3( a ), in the left hypomethylated region, the portion above the dotted line of the diseased tissue DNA is removed.

[0120] Step 1033: If the target differentially methylated region is a high-methylated region, remove the read data with a methylation state continuity score less than or equal to Smax in the second lesion sample read data set.

[0121] For example, as shown in FIG3( a ), in the high methylation region on the right, the portion below the dotted line of the diseased tissue DNA is removed.

[0122] Therefore, through the screening method of steps 1031 to 1033, on the one hand, the reads from various other cells (such as normal esophageal cells, immune cells, vascular endothelial cells, etc.) in the surgically removed lesion tissue (such as esophageal cancer tissue) can be filtered out as much as possible to improve the purity of the lesion reads in the training set. On the other hand, for example, when the methylation state of the ctDNA of the lesion cell overlaps with the DNA of other organs and tissues of the human body (such as the stomach, heart, etc.), the reads unique to the lesion tissue can be screened out to minimize the error of the neural network in judging the trace reads from other organs and tissues in the human blood as ctDNA.

[0123] In the above step 103, the step of calculating the methylation continuity score of each second lesion sample read data and the second control sample read data, and using the second control sample read data set as a reference and filtering out lesion-specific data from the second lesion sample read data set according to the methylation continuity score further includes:

[0124] Step 1034, respectively calculate the methylation continuity score of each second lesion sample read data and the methylation continuity score of each read data corresponding to the high-risk population in the second control sample read data set, and sort the methylation status continuity scores of all second control sample read data corresponding to the high-risk population from high to low.

[0125] For example, as shown in Figure 3(b), in the left hypomethylated region, all the read data from the patient's diseased tissue DNA (corresponding to the tumor tissue DNA on the horizontal axis), all the read data from the healthy population's cfDNA (corresponding to the healthy population's cfDNA on the horizontal axis), and the vertical axis is the distribution range of the MCS values ​​of all the read data (reads). The minimum value Smin of the methylation status continuity score in all the read data of the healthy population is marked in Figure 3(b).

[0126] For example, as shown in Figure 3(b), in the high methylation region on the right, all the read data from the patient's diseased tissue DNA (corresponding to the tumor tissue DNA on the horizontal axis), all the read data from the healthy population's cfDNA (corresponding to the healthy population's cfDNA on the horizontal axis), and the vertical axis is the distribution range of the MCS values ​​of all the read data (reads). The maximum value Smax of the methylation state continuity score in all the read data of the healthy population is marked in Figure 3(b).

[0127] Step 1035, if the target differentially methylated region is a low-methylated region, then remove a preset percentage (e.g., 5%, as shown in FIG. 3( b)) of the read data after the methylation status continuity score ranking in the second control sample read data set corresponding to the high-risk population, and record the minimum value of the methylation status continuity score of the remaining read data as Smin, and remove the read data with a methylation status continuity score greater than or equal to Smin in the second lesion sample read data set.

[0128] For example, as shown in FIG3( b ), in the left hypomethylated region, the portion above the dotted line of the diseased tissue DNA is removed.

[0129] In step 1036, if the target differentially methylated region is a high-methylated region, the read data with a methylation status continuity score ranking in the top preset percentage (e.g., 5%, as shown in FIG. 3( b)) in the second control sample read data set corresponding to the high-risk population is removed, and the maximum value of the methylation status continuity score of the remaining read data is recorded as Smax, and the read data with a methylation status continuity score less than or equal to Smax in the second lesion sample read data set is removed.

[0130] For example, as shown in FIG3( b ), in the high methylation region on the right, the portion below the dotted line of the diseased tissue DNA is removed.

[0131] Therefore, through the screening method of steps 1034 to 1036, some reads whose methylation patterns are similar to those of cfDNA of high-risk groups (such as hepatitis and cirrhosis DNA) can be further filtered out. The reason why 5% of the reads in cfDNA of high-risk groups are removed in advance is that it is not ruled out that a small amount of ctDNA is contained in the cfDNA of a small number of high-risk groups (.

[0132] After completing the reads screening and filtering in the previous step, the number of reads of the diseased tissue DNA of each DMR from each individual has changed. In order to screen out DMRs with more filtered reads in more individuals as much as possible. The lesion-specific data screening and processing method of the present invention also includes further screening of DMRs after step 1036. Therefore, in step 103, the second control sample read segment data set is used as a reference and the lesion-specific data is screened out from the second lesion sample read segment data set according to the methylation continuity score. It also includes:

[0133] Step 1037, after the removal operation, for each target differentially methylated region, according to a preset indicator function and the read data in all the second lesion sample read data sets corresponding to the target differentially methylated region, calculate an indicator value associated with the target differentially methylated region.

[0134] Exemplarily, the preset indicator function is a DMR universality score (DMR universality Score, DUS), and its calculation formula is:

[0135]

[0136] Among them, for each DMR, n represents the total number of individuals, t represents the ratio of the number of reads after filtering to the total number of reads before filtering, i represents the i-th individual, and d represents the ratio of individuals with t>0 to the total number of individuals.

[0137] Exemplarily, the value range of DUS is 0 to 1. DUS=0 means that the number of reads remaining after filtering the DMR in all individuals is 0; DUS=1 means that no reads of the DMR in all individuals have been filtered.

[0138] Step 1038: sort all index values ​​from large to small, and select the target differential methylation regions associated with the preset index values ​​with the highest sorting as the targeted differential methylation regions.

[0139] For example, the DMRs are arranged from large to small according to the DUS value, and the DMRs with preset index values ​​(for example, 200) ranked at the top are selected for subsequent analysis.

[0140] Step 1039 : determining the read data in the second lesion sample read data set after the removal operation corresponding to each targeted differentially methylated region as the lesion-specific data.

[0141] The above step 104 is described in detail below.

[0142] In the above step 104, the selected lesion-specific data is labeled and then the preset neural network is trained, so that the trained neural network can predict the probability of lesion occurrence based on the input samples.

[0143] For example, after the above filtering and screening steps, for example, all the lesion-specific data (reads) screened out in the above 200 DMRs are all labeled 1, and all the data (reads) from high-risk groups and healthy people's cfDNA are labeled 0. The labeled reads are used to train the neural network. The neural network structure can be a typical Transformer network.

[0144] Illustratively, after step 104 , the lesion-specific data screening and processing method according to the embodiment of the present invention further includes the following steps 105 and 106 .

[0145] Step 105, after the selected lesion-specific data are labeled and the preset neural network is trained, multiple DNA sample read data from the same target to be detected are input into the trained neural network to identify the probability of each read data being the lesion-specific data.

[0146] For example, all reads from the DMR of the circulating DNA (cfDNA) samples of the test set population are input into the trained neural network for identification. The neural network gives the probability that each read is derived from the circulating diseased DNA (ctDNA).

[0147] Step 106 , performing an estimation operation according to the probability corresponding to the DNA read data and a preset function to obtain the proportion of the pathological read data in the DNA read data of the target to be detected.

[0148] Exemplarily, the expression of the preset function is:

[0149]

[0150] in, It represents the proportion of diseased read data in the DNA read data of a target to be detected, p i The probability that the i-th read data output by the neural network is lesion-specific data, n represents the number of all read data from a target to be detected, and t represents a preset value.

[0151] For example, substitute t=0, 0.1%, 0.2%, 0.3%...99.9%, 100% into the right side of the above formula. In other words, divide the interval from 0% to 100% into 1000 equal parts at intervals of 0.1%, and take one of the values ​​of t each time. Find a value of t that maximizes the value of the above formula. This t value is the estimated ratio of ctDNA to cfDNA of the individual. The method can obtain the global optimal solution of t value with an accuracy of 0.1%.

[0152] For example, by setting a threshold, for the above calculated Individuals with values ​​greater than the threshold are judged as patients with lesions (eg, cancer patients); otherwise, they are judged as patients without lesions (eg, without cancer).

[0153] For example, a series of thresholds can be defined from 0% to 100% with a step length of 0.1%. Each threshold is used to make judgments on the population from the test set. As the threshold changes, the sensitivity, specificity and accuracy of the judgment on the test set will also change, so that the receiver operating characteristic curve (ROC) can be drawn and the area under the curve (AUC) value can be calculated.

[0154] By drawing the ROC curves in the results of the traditional method and the embodiment of the present invention and calculating the AUC value, the AUC values ​​between the two are compared. It is generally understood by those skilled in the art that the larger the AUC value, the better the method. The comparison results show that the denoising method implemented in the present application for screening lesion-specific data from high-throughput DNA methylation sequencing data is conducive to improving the recognition accuracy in subsequent analysis.

[0155] In summary, the embodiments of the present invention are beneficial to improving the training quality of the neural network and the final recognition accuracy by screening lesion-specific reads in DMRs and further screening DMRs.

[0156] The following is a description of the lesion-specific data screening and processing device provided by the present invention. The lesion-specific data screening and processing device described below and the lesion-specific data screening and processing method described above can be referenced to each other.

[0157] Figure 4 FIG. 1 is a schematic diagram of the structure of a lesion-specific data screening and processing device provided by an embodiment of the present invention. Figure 4 The embodiment of the present invention provides a lesion-specific data screening and processing device 400, the device 400 includes a pre-processing module 410, a search module 420, and a screening module 430. Among them,

[0158] The preprocessing module 410 is used to obtain lesion sample data and control sample data and preprocess the lesion sample data and the control sample data to obtain a first lesion sample read data set and a first control sample read data set, wherein each first lesion sample read data and each first control sample read data represents the base arrangement order on a DNA sequence and contains the methylation status information of each cytosine on the DNA sequence.

[0159] The search module 420 is used to search for at least one target differential methylation region between the two sets based on the first lesion sample read data set and the first control sample read data set, and obtain a second lesion sample read data set and a second control sample read data set representing each target differential methylation region.

[0160] The screening module 430 is used to calculate the methylation continuity score of each second lesion sample read data and the second control sample read data, and to screen out lesion-specific data from the second lesion sample read data set based on the methylation continuity score and the second control sample read data set as a reference.

[0161] Exemplarily, the device 400 further includes:

[0162] The training module 440 (not shown) is used to:

[0163] After the selected lesion-specific data are labeled, a preset neural network is trained so that the trained neural network can predict the probability of lesions based on input samples.

[0164] Exemplarily, the preprocessing module 410 is further used for:

[0165] Dividing the lesion sample data and the control sample data into a training set and a test set, wherein the sample data in the training set is used to screen out the lesion-specific data and to train a neural network, and the sample data in the test set is used to test the trained neural network;

[0166] The lesion sample data include DNA data of lesion tissues from a preset number of patients, and the control sample data include DNA data of non-lesion tissues adjacent to the lesion tissues of the preset number of patients, and DNA data of a preset number of healthy people and high-risk people.

[0167] Exemplarily, the preprocessing module 410 is further used for:

[0168] The DNA molecules of each sample are broken into DNA fragments of a preset length, and the DNA fragments are respectively subjected to enzymatic conversion to construct a DNA library;

[0169] Whole genome methylation sequencing is performed on the DNA library to obtain the first lesion sample read data set and the first control sample read data set.

[0170] Exemplarily, the preprocessing module 410 is further used for:

[0171] The performing whole genome methylation sequencing on the DNA library to obtain the first lesion sample read segment data set and the first control sample read segment data set comprises:

[0172] After whole-genome methylation sequencing of the DNA library, data filtering is performed on the sequencing results based on preset rules;

[0173] The filtered data is compared with a preset reference genome and the methylation status is identified to obtain the first diseased sample read data set and the first control sample read data set.

[0174] Exemplarily, the search module 420 is further configured to:

[0175] Searching for differentially methylated regions between first lesion sample read data and first control sample read data corresponding to the lesion tissues and non-lesion tissues of the preset number of patients, respectively, to obtain a first set;

[0176] Finding differentially methylated regions between the first lesion sample read data and the first control sample read data corresponding to the lesion tissues of the preset number of patients and the DNA data of the preset number of high-risk people, respectively, to obtain a second set;

[0177] Taking the intersection of the first set and the second set as the target differentially methylated region;

[0178] According to the target differentially methylated regions, a second lesion sample read data set and a second control sample read data set representing each target differentially methylated region are obtained.

[0179] Exemplarily, the screening module 430 is further used for:

[0180] Calculate the methylation continuity score of each second lesion sample read data and the methylation continuity score of each read data corresponding to the healthy population in the second control sample read data set, and mark the maximum and minimum values ​​of the methylation state continuity scores of all second control sample read data corresponding to the healthy population as Smax and Smin, respectively;

[0181] If the target differentially methylated region is a low-methylated region, removing the read data with a methylation state continuity score greater than or equal to Smin in the second lesion sample read data set;

[0182] If the target differentially methylated region is a high methylated region, the read data with a methylation state continuity score less than or equal to Smax in the second lesion sample read data set is removed.

[0183] Exemplarily, the screening module 430 is further used for:

[0184] Calculating the methylation continuity score of each second lesion sample read data and the methylation continuity score of each read data corresponding to the high-risk population in the second control sample read data set, and sorting the methylation status continuity scores of all second control sample read data corresponding to the high-risk population from high to low;

[0185] If the target differentially methylated region is a low-methylated region, a preset percentage of read data after ranking the methylation state continuity score in the second control sample read data set corresponding to the high-risk population is removed, and the minimum value of the methylation state continuity score of the remaining read data is recorded as Smin, and the read data with a methylation state continuity score greater than or equal to Smin in the second lesion sample read data set is removed;

[0186] If the target differentially methylated region is a high methylated region, the read data with a preset percentage of methylation status continuity scores ranked in the top of the second control sample read data set corresponding to the high-risk population are removed, and the maximum value of the methylation status continuity scores of the remaining read data is recorded as Smax, and the read data with a methylation status continuity score less than or equal to Smax in the second lesion sample read data set are removed.

[0187] Exemplarily, the screening module 430 is further used for:

[0188] After the removal operation, for each target differentially methylated region, according to a preset indicator function and the read data in all the second lesion sample read data sets corresponding to the target differentially methylated region, calculating an indicator value associated with the target differentially methylated region;

[0189] All index values ​​are sorted from large to small, and the target differential methylation regions associated with the preset index values ​​with the highest sorting are selected as the targeted differential methylation regions;

[0190] The read data in the second lesion sample read data set after the removal operation corresponding to each targeted differentially methylated region is determined as the lesion-specific data.

[0191] Exemplarily, the device 400 further includes:

[0192] An estimation module 450 (not shown in the figure) is used to:

[0193] After the selected lesion-specific data are labeled and a preset neural network is trained, DNA read data from the same target to be detected are input into the trained neural network to identify the probability of whether each read data is the lesion-specific data;

[0194] An estimation operation is performed according to the probability corresponding to the DNA read segment data and a preset function to obtain the proportion of the pathological read segment data in the DNA read segment data of the target to be detected.

[0195] Exemplarily, the expression of the preset function is:

[0196]

[0197] in, It represents the proportion of diseased read data in the DNA read data of a target to be detected, p i The probability that the i-th read data output by the neural network is lesion-specific data, n represents the number of all read data from a target to be detected, and t represents a preset value.

[0198] Figure 5 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 5As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530 and a communication bus 540, wherein the processor 510, the communication interface 520 and the memory 530 communicate with each other via the communication bus 540. The processor 510 may call the logic instructions in the memory 530 to execute the lesion-specific data screening and processing method, and the method includes:

[0199] Acquire lesion sample data and control sample data and preprocess the lesion sample data and the control sample data to obtain a first lesion sample read segment data set and a first control sample read segment data set, wherein each first lesion sample read segment data and each first control sample read segment data represent the base arrangement order on a DNA sequence and contain methylation state information of each cytosine on the DNA sequence;

[0200] Based on the first lesion sample read data set and the first control sample read data set, searching for at least one target differential methylation region between the two sets, and obtaining a second lesion sample read data set and a second control sample read data set representing each target differential methylation region;

[0201] The methylation continuity score of each second lesion sample read data and the second control sample read data is calculated, and the lesion-specific data is screened out from the second lesion sample read data set with the second control sample read data set as a reference and according to the methylation continuity score.

[0202] In addition, the logic instructions in the above-mentioned memory 530 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0203] On the other hand, the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the lesion-specific data screening and processing methods provided by the above-mentioned methods.

[0204] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the above-mentioned lesion-specific data screening and processing methods.

[0205] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0206] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0207] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for screening and processing lesion-specific data, characterized in that: The method comprises: Acquire lesion sample data and control sample data, and preprocess the lesion sample data and the control sample data to obtain a first lesion sample read data set and a first control sample read data set, wherein each first lesion sample read data and each first control sample read data represent the base arrangement order on a DNA sequence and contain methylation state information of each cytosine on the DNA sequence; Based on the first lesion sample read data set and the first control sample read data set, searching for at least one target differential methylation region between the two sets, and obtaining a second lesion sample read data set and a second control sample read data set representing each target differential methylation region; Calculating the methylation continuity score of each second lesion sample read data and the second control sample read data, and screening out lesion-specific data from the second lesion sample read data set based on the methylation continuity score and taking the second control sample read data set as a reference, specifically comprising: Calculate the methylation continuity score of each second lesion sample read data and the methylation continuity score of each read data corresponding to the healthy population in the second control sample read data set, and calculate the maximum and minimum values ​​of the methylation state continuity scores of all second control sample read data corresponding to the healthy population; If the target differential methylation region is a low methylation region, then the read data with a methylation state continuity score greater than or equal to the minimum value in the second lesion sample read data set is removed; if the target differential methylation region is a high methylation region, then the read data with a methylation state continuity score less than or equal to the maximum value in the second lesion sample read data set is removed; After the removal operation, for each target differential methylation region, the index value associated with the target differential methylation region is calculated according to a preset index function and the read data in all second lesion sample read data sets corresponding to the target differential methylation region; all index values ​​are sorted from large to small, and the target differential methylation region associated with the preset index values ​​with the highest sorting order is selected as the targeted differential methylation region; the read data in the second lesion sample read data set after the removal operation corresponding to each targeted differential methylation region is determined as lesion-specific data.

2. The lesion-specific data screening and processing method according to claim 1, characterized in that: Acquiring lesion sample data and control sample data and preprocessing the lesion sample data and the control sample data to obtain a first lesion sample read segment data set and a first control sample read segment data set also includes: The DNA molecules of each sample are broken into DNA fragments of a preset length, and the DNA fragments are respectively converted by enzymatic or salt methods to construct a DNA library; Whole genome methylation sequencing is performed on the DNA library to obtain the first lesion sample read data set and the first control sample read data set.

3. The lesion-specific data screening and processing method according to claim 2, characterized in that: The performing whole genome methylation sequencing on the DNA library to obtain the first lesion sample read segment data set and the first control sample read segment data set comprises: After whole-genome methylation sequencing of the DNA library, data filtering is performed on the sequencing results based on preset rules; The filtered data is compared with a preset reference genome and the methylation status is identified to obtain the first diseased sample read data set and the first control sample read data set.

4. The lesion-specific data screening and processing method according to claim 3, characterized in that: The step of searching for at least one target differentially methylated region between the first lesion sample read data set and the first control sample read data set based on the two sets, and obtaining a second lesion sample read data set and a second control sample read data set representing each target differentially methylated region comprises: Searching for differentially methylated regions between first lesion sample read data and first control sample read data corresponding to the lesion tissues and non-lesion tissues of the preset number of patients, respectively, to obtain a first set; Finding differentially methylated regions between the first lesion sample read data and the first control sample read data corresponding to the lesion tissues of the preset number of patients and the DNA data of the preset number of high-risk people, respectively, to obtain a second set; Taking the intersection of the first set and the second set as the target differentially methylated region; According to the target differentially methylated regions, a second lesion sample read data set and a second control sample read data set representing each target differentially methylated region are obtained.

5. The lesion-specific data screening and processing method according to claim 1, characterized in that: The step of calculating the methylation continuity score of each second lesion sample read data and the second control sample read data, and taking the second control sample read data set as a reference and filtering out lesion-specific data from the second lesion sample read data set according to the methylation continuity score further comprises: Calculating the methylation continuity score of each second lesion sample read data and the methylation continuity score of each read data corresponding to the high-risk population in the second control sample read data set, and sorting the methylation status continuity scores of all second control sample read data corresponding to the high-risk population from high to low; If the target differentially methylated region is a low-methylated region, a preset percentage of read data after ranking the methylation state continuity score in the second control sample read data set corresponding to the high-risk population is removed, and the minimum value of the methylation state continuity score of the remaining read data is recorded as Smin, and the read data with a methylation state continuity score greater than or equal to Smin in the second lesion sample read data set is removed; If the target differentially methylated region is a high methylated region, the read data with a preset percentage of methylation status continuity scores ranked in the top of the second control sample read data set corresponding to the high-risk population are removed, and the maximum value of the methylation status continuity scores of the remaining read data is recorded as Smax, and the read data with a methylation status continuity score less than or equal to Smax in the second lesion sample read data set are removed.

6. A lesion-specific data screening and processing device, characterized in that: The device comprises: A preprocessing module, used to obtain lesion sample data and control sample data and preprocess the lesion sample data and the control sample data to obtain a first lesion sample read segment data set and a first control sample read segment data set, wherein each first lesion sample read segment data and each first control sample read segment data represent the base arrangement order on a DNA sequence and contain methylation status information of each cytosine on the DNA sequence; A search module, configured to search for at least one target differential methylation region between the first lesion sample read data set and the first control sample read data set based on the first lesion sample read data set and the first control sample read data set, and obtain a second lesion sample read data set and a second control sample read data set representing each target differential methylation region; A screening module is used to calculate the methylation continuity score of each second lesion sample read data and the second control sample read data, and to screen out lesion-specific data from the second lesion sample read data set based on the methylation continuity score and the reference of the second control sample read data set, specifically comprising: Calculate the methylation continuity score of each second lesion sample read data and the methylation continuity score of each read data corresponding to the healthy population in the second control sample read data set, and calculate the maximum and minimum values ​​of the methylation state continuity scores of all second control sample read data corresponding to the healthy population; If the target differential methylation region is a low methylation region, then the read data with a methylation state continuity score greater than or equal to the minimum value in the second lesion sample read data set is removed; if the target differential methylation region is a high methylation region, then the read data with a methylation state continuity score less than or equal to the maximum value in the second lesion sample read data set is removed; After the removal operation, for each target differential methylation region, the index value associated with the target differential methylation region is calculated according to a preset index function and the read data in all second lesion sample read data sets corresponding to the target differential methylation region; all index values ​​are sorted from large to small, and the target differential methylation region associated with the preset index values ​​with the highest sorting order is selected as the targeted differential methylation region; the read data in the second lesion sample read data set after the removal operation corresponding to each targeted differential methylation region is determined as lesion-specific data.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the lesion-specific data screening and processing method according to any one of claims 1 to 5 is implemented.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the lesion-specific data screening and processing method according to any one of claims 1 to 5 is implemented.