Method and application of filtering false positive mutations caused by DNA damage in NGS sequencing

Abstract

The present invention discloses a method and application for filtering false positive mutations of DNA damage in NGS sequencing, and belongs to the field of gene detection technology. The method includes: S1, constructing a baseline based on the distribution of mutant subtypes of normal samples; S2, performing variation detection on the sample to be tested, judging the difference between the mutant subtype corresponding to the sample to be tested and the baseline by binomial test or negative binomial test, and judging whether the sample to be tested has DNA damage according to the difference; S3, extracting SNV mutations with VAF≤5% for samples with DNA damage, and determining the baseline error rate of the corresponding mutant subtype at the benchmark sequencing depth; S4, calculating the DNA damage significance p value of the site, and judging whether the corresponding site is a true mutation site or a false positive site based on the p value. The present invention can not limit the variant allele frequency of DNA damage, and can be applied to more technical platforms at the same time, without being restricted by single-end or double-end sequencing.

Description

Technical Field

[0001] The present invention belongs to the technical field of gene detection, and specifically relates to a method and application of filtering false positive mutations caused by DNA damage in NGS (Next-Generation Sequencing, also known as second-generation sequencing) sequencing. Background Art

[0002] Formalin-fixed paraffin-embedded samples (FFPE) are often used to preserve patient tissue samples in clinical pathology, tumor gene testing, and medical scientific research because they are easy to preserve for a long time. When using second-generation sequencing technology for precision treatment, these tissue samples are largely relied on. However, some sequence mutations detected in FFPE DNA samples may be caused by damage during sample processing, which mainly comes from the following situations:

[0003] 1) Oxidative damage

[0004] Reactive oxygen species (ROS) can oxidize guanine (G) to 8-oxoguanine (8-oxoG). 8-oxoG no longer pairs with C, but pairs with A. Therefore, after two replications, the G:C pairing is converted to A:T pairing, resulting in a false positive mutation (G:C / T:A G->T / C->A mutation), such as Figure 1 shown.

[0005] 2) DNA damage caused by FFPE treatment

[0006] a. Cross-linking: cross-linking occurs between macromolecules (such as proteins and nucleic acids). The reason why formaldehyde causes DNA damage is mainly due to its carbonyl electrophilicity and small steric hindrance, which makes it easy to cross-link with nucleic acids and proteins. In vitro, formaldehyde first reacts with free amino groups on proteins or nucleic acids to form unstable hydroxymethyl adducts, and then further reacts with other nucleic acids or proteins to form stable cross-links. This causes local denaturation of DNA and affects the quality of nucleic acid extraction.

[0007] b. Base conversion: There are a large number of C>U or C>T mutations in FFPE DNA samples. In vivo, cytosine spontaneously hydrolyzes and deaminates to uracil, with a frequency of about 190 times per cell per day. It can be restored to cytosine through base excision repair, but it cannot be repaired in vitro. Therefore, a large number of C:G / T:A mutation false positives (C->T / G->A) will be found in the sequencing data. Figure 2 shown.

[0008] c. Base shedding: The abasic site is formed when the DNA molecule loses a purine base or a pyrimidine base. When the tissue is fixed in formalin, the formaldehyde in the formalin is easily oxidized into formic acid in the air, and the formic acid lowers the pH value of formalin. The N-glycosidic bond between the purine base and the sugar backbone is easily hydrolyzed at low pH, causing the DNA base to fall off. In addition, the abasic site can be spontaneously cut through the B-elimination reaction, resulting in DNA chain breakage.

[0009] 3) DNA fragmentation

[0010] The degree of fragmentation of FFPE DNA increases with the extension of storage time and the decrease of pH value of formalin used in fixation. Compared with fresh FFPE DNA, the success rate of PCR is lower for FFPE DNA that has been stored for a long time, indicating that DNA fragmentation may have been occurring during the storage of the sample.

[0011] Based on the causes of DNA damage and the principles of NGS sequencing, it can be found that this false positive site has chain preference, that is, the mutation only exists on the positive or negative chain of the DNA molecule, and the reads obtained in the NGS sequencing process show the characteristics of F1R2 or F2R1.

[0012] Common filtering methods currently used include filtering initial candidate variants based on statistical laws. For example, Yost et al. used a binomial test to test the allele frequency of each variant and its corresponding mutation type in the global mismatch rate to control formalin-induced deamination, and then removed insignificant allele frequencies. Similarly, Kerick et al. used a cutoff value for the sequencing depth of the variant site for filtering. The FIREVAT algorithm uses a series of filtering parameters, including VAF, alignment quality, and the depth of the reference sequence and allele sequence.

[0013] In addition, there are some filtering algorithms used to identify false positive variants in low-quality or FFPE sequencing data. For example, the cisCall algorithm applies continuous filters, such as for clustered low VAF mutations, if it is higher than the expected error rate of mutations occurring within the clustered interval, it will be filtered through internal statistical calculations. There are also Bayesian variant detection algorithms that, in addition to applying common quality filters, also combine prior information about specific cancer-related mutations. Another algorithm, LoLoPicker, uses mutations within the FFPE sample panel to assess the background error rate of specific sites, and then incorporates it into hypothesis testing to identify true and false variants. The mutation detection software developed by Pisces includes a model that recalibrates the quality of variants based on the deviation from the average mutation rate of each possible mutation type, thereby minimizing thermal damage and FFPE deamination.

[0014] FFPE-related deamination, like false positives such as oxidative damage to DNA, usually only occurs on a single strand of the original DNA template, resulting in a directional deviation between reads1 and reads2 during double-end sequencing. Based on this chain preference, DNA damage can be effectively measured. This method of calculating chain imbalance has been adopted by multiple software, such as the LearnReadOrientationModel module of GATK4, which is used to filter false positive mutations that only appear on a single strand during sequencing. This tool mainly uses a Bayesian statistical model to calculate the chain preference generated by single nucleotide substitutions in every three bases. In addition, the SOBDetector software also incorporates chain preference parameters, but instead of direct filtering, it calculates the chain deviation score for each variant and adds the value to the mutation result file (vcf) for manual screening.

[0015] However, some of the above algorithms will discard variants with VAF <5% or only evaluate performance on VAF>5%. Because, the general consensus is to discard C:G>T:A variants when the variant allele frequency (VAF) is <5% or <10%, because FFPE-related false positives usually only exist at low frequencies. However, this limitation is not ideal because it prevents the detection of potentially clinically relevant low-frequency variants and limits the use of low tumor content samples. In addition, FFPE-related false positives have also been observed at frequencies >10%.

[0016] Since DNA damage mutations are located on a single strand of the original DNA template, current software or algorithms incorporate the preference of the DNA template into calculations or feature training, which is not applicable to single-end sequencing NGS data. From the current literature, most of the test data are capture data, and there is relatively little data on amplicon. The performance of single-end sequencing amplicon data on DNA damage samples has not yet been seen. Summary of the invention

[0017] The present invention aims to solve one of the technical problems in the above-mentioned related art at least to a certain extent.

[0018] To this end, the purpose of the present invention is to provide a method and application for filtering false positive mutations caused by DNA damage in NGS sequencing, which can not limit the variant allele frequency (VAF) of DNA damage, and can be applied to more technical platforms without being restricted by single-end or double-end sequencing.

[0019] In order to solve the above-mentioned technical problems, the present invention is achieved as follows:

[0020] An embodiment of the present invention provides a method for filtering false positive mutations of DNA damage in NGS sequencing, the method comprising:

[0021] S1. Build a baseline based on the mutation subtype distribution of normal samples;

[0022] S2. Perform mutation detection on the sample to be tested, and based on the detection results, determine the difference between the mutant subtype corresponding to the sample to be tested and the baseline by binomial test or negative binomial test, and determine whether the sample to be tested has DNA damage according to the difference;

[0023] S3. For samples with DNA damage, extract SNV mutations with VAF ≤ 5% and determine the baseline error rate of the corresponding mutation subtypes at the baseline sequencing depth;

[0024] S4. Calculate the DNA damage significance p value of the site according to the benchmark error rate, and determine whether the corresponding site is a true mutation site or a false positive site based on the DNA damage significance p value, thereby achieving filtering.

[0025] In addition, the method for filtering false positive mutations caused by DNA damage in NGS sequencing according to the present invention may also have the following additional technical features:

[0026] In some embodiments, the contents of constructing a baseline based on the mutant subtype distribution of a normal sample in step S1 include:

[0027] Several normal samples without DNA damage were selected for variation detection, and the original SNV results of variation detection were counted to obtain the distribution of the mutation subtype to be tested in normal samples, and a baseline was constructed based on the distribution.

[0028] In some implementations, the content of step S2 includes:

[0029] The frequencies of different mutation subtypes to be tested in the variation detection results of the tested samples are counted, and a one-sided binomial test or a negative binomial test is performed based on the baseline to calculate the p-value; the obtained p-value is compared with the preset threshold; when the p-value is not greater than the preset threshold, and the number of corresponding mutation subtypes in the tested sample is not less than the preset reference value, it is determined that DNA damage exists in the sample, otherwise it is determined that there is no DNA damage.

[0030] In some of the embodiments, the preset threshold corresponding to the p-value is 0.01.

[0031] In some of the embodiments, the benchmark sequencing depth is obtained by calculating the median of the sequencing depths of all SNVs in the corresponding sample as the benchmark sequencing depth.

[0032] In some of the embodiments, in step S3, the sequencing depths, supporting sequencing base sequences, and VAFs of all mutation subtypes to be tested are counted, SNV mutations with VAF ≤ 5% are extracted based on the statistical information, and the 95% quantile is selected as the baseline error rate of the mutation subtype to be tested in the sample at the baseline sequencing depth.

[0033] In some of these embodiments, the baseline error rate is negatively correlated to the power of log(sequencing depth).

[0034] In some embodiments, the specific relationship between the benchmark error rate and sequencing depth is as follows:

[0035] \in =\frac {e*\left [ {{log}_{10}\left ( {MedianDepth} \right )} \right ]^{3}} {\left [ {{log}_{10}\left ( {Depth} \right )} \right ]^{3}} .

[0036] In some of the implementations, it is also necessary to preset an ultra-high depth cutoff value:

[0037] The sequencing depth is compared with the ultra-high depth cutoff value. When the sequencing depth is greater than the ultra-high depth cutoff value, the error rate no longer decreases with the increase of sequencing depth. The error rates of sites with a depth higher than the cutoff value use the error rate corresponding to the cutoff value depth and remain unchanged.

[0038] An embodiment of the present invention also provides an application of the method for filtering false positive mutations of DNA damage in NGS sequencing as described in any of the above items, wherein the method is applied in single-end sequencing SE or double-end sequencing PE.

[0039] Compared with the prior art, the present invention has at least the following beneficial effects:

[0040] In the embodiments of the present invention, the method for filtering false positive mutations caused by DNA damage in NGS sequencing can solve the current dependence on chain preference in the identification and filtering of false positive mutations caused by DNA damage. It can also be applied to mutation sites of different depths and VAFs, as well as data from different sequencing platforms and sequencing methods, providing support for accurate detection of tumor genes.

[0041] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is an example diagram of DNA oxidative damage 8-oxoG in the prior art;

[0043] Figure 2 This is an example of DNA damage caused by FFPE treatment in the prior art; specifically, DNA cytosine deamination leads to the conversion of cytosine to uracil, resulting in C:G / T:A (G->A / C->T mutation);

[0044] Figure 3 A flowchart of a method for filtering false positive mutations caused by DNA damage in NGS sequencing disclosed in one embodiment of the present invention. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments 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.

[0046] The embodiments of the present invention are described in detail below through specific embodiments and application scenarios in conjunction with the accompanying drawings.

[0047] See also Figure 1 As shown, in some embodiments of the present invention, a method for filtering false positive mutations of DNA damage in NGS sequencing is provided, which can overcome the current technical drawbacks, does not limit the variant allele frequency (VAF) of DNA damage, and can be applied to more technical platforms, is not limited by single-end or double-end sequencing, and has broad application prospects in accurate tumor detection.

[0048] In some embodiments of the present invention, the method for filtering false positive mutations of DNA damage in NGS sequencing specifically includes:

[0049] First, a baseline is constructed based on the mutation distribution of 12 subtypes in normal samples, and a Binomial test or a Negative Binomial test is used to determine whether the subtype corresponding to the sample to be tested is significantly different from the baseline. The cutoff of DNA damage samples is determined by a binomial p-value ≤ 0.01. In conventional DNA damage judgment or filtering software, the GATK CollectOxoGMetrics or FilterByOrientationBias module is usually used to calculate the TOTAL_QSCORE value of the 12 subtypes in the sample. The lower the score, the greater the possibility of artificial changes. At the same time, the F1R2 and F2R1 values ​​are used to determine whether the mutation is a false positive mutation caused by DNA damage. Therefore, this method does not directly determine the DNA damage of the sample, and the subsequent site determination also needs to rely on chain preference (whether the values ​​of the support reads of the mutation on F1R2 or F2R1 are biased) for filtering. Among them, GATK refers to Genome Analysis Toolkit, genome analysis toolkit; CollectOxoGMetrics refers to the module for collecting oxidized guanine (OxoG) metrics; FilterByOrientationBias refers to the module for filtering by orientation bias.

[0050] Secondly, after determining that the sample to be tested is a DNA-damaged sample, the median depth (Median Depth) and the 95% percentile allele frequency (Vaf) of the 12 subtypes in the sample to be tested are statistically analyzed, and the depth and frequency are used as the benchmark depth and benchmark error rate of the corresponding mutation type. For each potential DNA damage false positive mutation type (G->A / C->T; G->T / CA) site in the sample to be tested, according to the actual sequencing depth (Depth) of each site, the dynamic error rate ϵ (see Formula 1) is used as the background error probability of the site, so as to enter the next step of site authenticity assessment. When the Depth increases, a lower error rate is used to avoid filtering out high-frequency real mutations. At the same time, in order to avoid the error rate dropping to an unrealistically low value at ultra-high Depth, the median depth of the sample to be tested is used as a cutoff value. When the depth is greater than this value, the error rate no longer decreases with the increase of depth. The dynamic error rate will continue to increase with the decrease of depth, which is also consistent with the background Vaf of DNA damage sites being higher at low depths. The dynamic error rate can effectively evaluate the true performance of DNA damage sites at ultra-low depth (<100X) or ultra-high depth (≥5000X). In current software or algorithms, although binomial distribution or other algorithms are used to evaluate DNA damage sites, the test data depth of tissue samples is relatively low, and the test effect of DNA damage at ultra-high depth is unknown. The dynamic error rate in the present invention can make up for this technical shortcoming.

[0051] Finally, for each site to be tested in the sample to be tested, the current dynamic error rate ϵ and sequencing depth (Depth) are used to obtain the p-value (see Formula 2) as the significance value p-damage of the DNA damage at the site. If p-damage>0.001, it means that there is no significant difference in the background error rate of the site and the mutation type corresponding to the site in the sample, and the site will be filtered out. In conventional software or algorithms, chain preference is used as an important condition to determine whether the variation is caused by DNA damage. Therefore, it is incompatible with single-end sequencing data, thereby affecting the effect of algorithm filtering. The present invention does not need to determine whether the data belongs to single-end sequencing or double-end sequencing, and can be included in the calculation. In addition, some software or algorithms will limit the sites of DNA damage to within 5%. The present invention can still exclude potential false-positive sites of DNA damage for low frequencies.

[0052] In real tumor clinical samples, the present invention detects samples that have been stored in FFPE for more than 2 years or more than 5 years. The calculated p-sample and the degree of DNA degradation detected by the experimental method show a high linear correlation. The larger the value of p-sample, the higher the degree of degradation of the sample and the more serious the DNA damage. The p-sample value in the present invention can effectively reduce the experimental steps and is more sensitive than the experimental test results. By filtering the mutations after the present invention, the false positive mutation sites can be greatly reduced, which is suitable for low-depth or ultra-high-depth sample detection. In addition, the present invention is simple to operate and can be applied to a variety of platforms and a variety of sequencing methods (single-end sequencing SE or double-end sequencing PE).

[0053] Embodiment 1:

[0054] In order to accurately filter out base mutations caused by DNA damage, the present invention designs a filtering method for false positive mutations caused by DNA damage. The method first identifies samples with DNA damage. This step can distinguish DNA damage samples from normal samples and avoid accidental damage to low-frequency mutations in normal samples during filtering; then the previously identified DNA damage samples are filtered, and the p-value of the site to be tested is calculated to determine whether the site is introduced by DNA damage based on the p-value. Specifically:

[0055] a. Use the original SNV (single nucleotide variation) results after mutation detection of several normal DNA samples (manually judged to have no DNA damage) to obtain the distribution of the mutation subtype to be tested in normal samples and build the corresponding baseline.

[0056] b. Perform the same processing on the original SNV results after the variation detection of the sample to be tested, and count the frequencies of different mutation subtypes to be tested, and perform a one-sided Binomial test or Negative Binomial test on the baseline to calculate the p-value. When the p-value ≤ 0.01, it is considered that there is DNA damage in the sample; if the number of corresponding mutation subtypes in the sample to be tested is less than the reference value, it is not considered to be DNA damage even if the p-value is significant.

[0057] c. For samples judged to have DNA damage, the median sequencing depth of all SNVs in the sample is counted as the benchmark sequencing depth; the sequencing depth, support reads (supporting sequencing base sequences), VAF and other information of all mutation subtypes to be tested are counted, and SNV mutations with VAF ≤ 5% are extracted, and the 95% quantile is selected as the benchmark error rate of the mutation subtype to be tested in the sample at the benchmark sequencing depth. .

[0058] Site depth dynamic error rate , the error rate of DNA damage type sites is different from the constant error rate of other error types. Its error rate increases with decreasing depth:

[0059] \in =\frac {e*\left [ {{log}_{10}\left ( {MedianDepth} \right )} \right ]^{3}} {\left [ {{log}_{10}\left ( {Depth} \right )} \right ]^{3}} (1)

[0060] In the formula, e is the mutation type to be tested, such as the vaf of C->T / G->A at the 90% quantile in the sample to be tested, which is used as the baseline error rate;

[0061] MedianDepth is the median depth of the sample;

[0062] Depth is the depth of the site to be measured.

[0063] Formula (1) works better in some implementation cases. Although other formulas can be used in theory, after a large number of experiments and comparisons, it has been verified that formula (1) is the best for filtering samples with low depth of DNA damage.

[0064] d. Each sequencing depth corresponds to an error rate. The detailed formula is shown in formula (1). The error rate calculated by this formula is different from the fixed error rate. It decreases exponentially with the increase of sequencing depth. When the sequencing depth is low, a larger error rate can be obtained to filter out mutations introduced by DNA damage; when the sequencing depth is very high, a lower error rate can be obtained to avoid filtering out high-frequency real mutations. In addition, in order to avoid the error rate falling to an extremely low level at an ultra-high sequencing depth, the present invention sets a cutoff value. When the sequencing depth is greater than the cutoff value, the error rate no longer decreases with the increase of sequencing depth, and the error rate corresponding to the cutoff value depth is used as the error rate of the current sequencing depth. The error rate corresponding to the cutoff value depth is also calculated by formula (1).

[0065] e. Use the above error rate to calculate the significance p-value of DNA damage at the site, damage_p-value, as shown in formula (2). Filter the mutation subtypes to be tested based on damage_p-value. When damage_p-value≤0.001, the site is considered to be a true mutation, otherwise it is considered to be a false positive site introduced by DNA damage.

[0066] Calculation of the significance p-value of DNA damage at the site, that is, calculating the p-value of each C->T / G->A mutation site by performing a Binomial test (or Negative Binomial test) at the current error rate and depth:

[0067] (2)

[0068] In the formula, SR is the number of reads supporting the mutation (Support Reads);

[0069] D is Depth, the number of all reads at the site (including reads that do not support mutations, that is, reference alleles);

[0070] is the error rate.

[0071] The method for filtering DNA damage false positive mutations in NGS sequencing of the present invention. First, potential DNA damage samples can be clearly determined based on baseline samples and the calculated p-sample, and the severity of DNA damage in the sample can be determined according to the size of the p-sample value (as shown in Table 1). In the experimental process, the N / Q ratio is generally used as the standard for judging the quality of DNA samples. When 0 < N / Q ratio ≤ 3.5, the integrity of the sample DNA is good; when 3.5 < N / Q ratio ≤ 6.5, the integrity of the sample DNA is relatively good; when the N / Q ratio > 6.5, the integrity of the sample DNA is poor. It can be seen from the table that as the N and Q ratios increase continuously, the value of p-sample becomes lower and lower, indicating that the sample quality becomes worse and worse. Therefore, the present invention can assist in experimental determination of sample quality, serve as an important parameter for sample quality control, reduce experimental steps, and save experimental costs. Secondly, it can solve the dependence on strand bias in the identification and filtering of DNA damage false positive mutations. Some FFPE samples have accumulated C>T and G>A damage mutations in some fragments due to long-term storage. At the same time, due to the existence of these mutations, the PCR amplification efficiency of such fragments is low during the library construction process before sequencing, resulting in low measured depth. Therefore, the sequencing data in these regions show the characteristics of low depth and high variant allele frequency (VAF). In software of the same type such as GATK, for single-end sequencing data in DNA damage samples, due to the software's dependence on strand bias, it is impossible to effectively filter them out. However, through comparative tests, the method of the present invention can effectively filter out such errors. Moreover, the present invention can also be applied to mutation sites with different depths and VAFs. For low-quality samples, while the degradation is relatively serious, the DNA damage is also very serious. The amplification efficiency of this type of sample is relatively low, and it is difficult to distinguish the true and false mutations when the sample depth is insufficient. However, the present invention can perform false positive filtering on this type of sample, increasing the detection and medication probability of patients. Finally, NGS detection is widely used in precision medicine. In addition to the common capture method, the amplicon amplification method is widely used because of its fast detection speed, usually producing results within 24 hours. However, this detection method is different on the MGI platform and multiple platforms of Illumina due to differences in sequencing error rates. The present invention can be compatible with the sequencing data of multiple platforms and different sequencing methods at the same time, providing support for the wide application of precise tumor gene detection.

[0072]

[0073] For parts not described in detail in the present invention, reference can be made to the existing technologies in the field or the well-known technologies to those skilled in the art.

[0074] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are within the protection of the present invention.

Claims

1. A method for filtering false positive mutations of DNA damage in NGS sequencing, characterized in that: The method comprises: S1. Build a baseline based on the mutation subtype distribution of normal samples; S2. Perform mutation detection on the sample to be tested, and based on the test results, determine the difference between the mutant subtype corresponding to the sample to be tested and the baseline by binomial test or negative binomial test, and determine whether the sample to be tested has DNA damage according to the difference; S3. For samples with DNA damage, extract SNV mutations with VAF ≤ 5% and determine the baseline error rate of the corresponding mutation subtypes at the baseline sequencing depth; S4. Calculate the DNA damage significance p value of the site according to the benchmark error rate, and determine whether the corresponding site is a true mutation site or a false positive site based on the DNA damage significance p value, thereby achieving filtering.

2. The method for filtering false positive mutations of DNA damage in NGS sequencing according to claim 1, characterized in that: The contents of constructing a baseline based on the mutation subtype distribution of normal samples in step S1 include: Several normal samples without DNA damage were selected for variation detection, and the original SNV results of variation detection were counted to obtain the distribution of the mutation subtype to be tested in normal samples, and a baseline was constructed based on the distribution.

3. The method for filtering false positive mutations of DNA damage in NGS sequencing according to claim 1, characterized in that: The content of step S2 includes: The frequencies of different mutation subtypes to be tested in the variation detection results of the tested samples are counted, and a one-sided binomial test or a negative binomial test is performed based on the baseline to calculate the p-value; the obtained p-value is compared with the preset threshold; when the p-value is not greater than the preset threshold, and the number of corresponding mutation subtypes in the tested sample is not less than the preset reference value, it is determined that DNA damage exists in the sample, otherwise it is determined that there is no DNA damage.

4. The method for filtering false positive mutations of DNA damage in NGS sequencing according to claim 3, characterized in that: The preset threshold corresponding to the p-value is 0.

01.

5. The method for filtering false positive mutations of DNA damage in NGS sequencing according to claim 1, characterized in that: The benchmark sequencing depth is obtained by calculating the median of the sequencing depths of all SNVs in the corresponding sample as the benchmark sequencing depth.

6. The method for filtering false positive mutations of DNA damage in NGS sequencing according to claim 1, characterized in that: In step S3, the sequencing depth, supporting sequencing base sequences, and VAF of all mutation subtypes to be tested are counted, and SNV mutations with VAF ≤ 5% are extracted based on the statistical information, and the 95% quantile is selected as the baseline error rate of the mutation subtype to be tested in the sample at the baseline sequencing depth.

7. The method for filtering false positive mutations of DNA damage in NGS sequencing according to claim 1, characterized in that: The baseline error rate is negatively correlated with log(sequencing depth).

8. The method for filtering false positive mutations of DNA damage in NGS sequencing according to claim 1, characterized in that: The relationship between the benchmark error rate and sequencing depth is: \in =\frac {e*\left [ {{log}_{10}\left ( {MedianDepth} \right )} \right ]^{3}} {\left [ {{log}_{10}\left ( {Depth} \right )} \right ]^{3}} (1) 。 9. The method for filtering false positive mutations of DNA damage in NGS sequencing according to claim 1, characterized in that: Preset super-high depth cutoff value; The sequencing depth is compared with the ultra-high depth cutoff value. When the sequencing depth is greater than the ultra-high depth cutoff value, the error rate corresponding to the cutoff value depth is used as the error rate of the current sequencing depth.

10. An application of the method for filtering false positive mutations of DNA damage in NGS sequencing according to any one of claims 1 to 9, characterized in that: The method is applied in single-end sequencing SE or double-end sequencing PE.

Images