Method for judging microsatellite instability based on fluorescent PCR capillary electrophoresis and application thereof

Abstract

The invention discloses a method for judging microsatellite instability based on fluorescent PCR capillary electrophoresis, which comprises the following steps of: acquiring a fluorescent PCR capillary electrophoresis detection result of a to-be-detected site and fluorescent PCR capillary electrophoresis of a microsatellite stable site, and converting the fluorescent PCR capillary electrophoresis detection result and the fluorescent PCR capillary electrophoresis detection result into matrix data through data processing; obtaining the matrix transverse displacement, the number of newly increased peaks and the number of peak increasing peaks when the minimum error fraction of the to-be-detected site is obtained; and inputting the matrix transverse displacement at the minimum error score, the number of newly increased peaks and the number of peak increasing peaks into a trained support vector machine to obtain a microsatellite instability result of the to-be-measured site. According to the method, the microsatellite instability of the to-be-detected site can be automatically judged based on the fluorescent PCR capillary electrophoresis of the to-be-detected site, subjective errors of traditional microsatellite instability judgment are eliminated, and then false positive and false negative are reduced.

Description

Technical Field

[0001] The present invention relates to the field of biomedicine technology, and in particular to a method for judging microsatellite instability based on fluorescence PCR capillary electrophoresis and an application thereof. Background Art

[0002] Microsatellite (MS) refers to a DNA sequence in the cell genome that is repeated in tandem with a few nucleotides (mostly 1 to 6); members of the mismatch repair (MMR) system include proteins such as MLH1, MSH2, MSH6 and PMS2. Occasionally, small DNA mismatch errors occur during DNA replication, which can be recognized and cut by these proteins, and then synthesized into new chains for repair; when the function of DNA mismatch repair genes is abnormal (dMMR), the replication errors of microsatellites cannot be corrected and continue to accumulate, causing changes in the length or base composition of microsatellite sequences, which becomes microsatellite instability (MSI).

[0003] MSI can be divided into three categories according to the degree of instability: microsatellite high instability (MSI-high, MSI-H), microsatellite low instability (MSI-low, MSI-L) and microsatellite stability (MSS); it can also be divided into two categories: microsatellite high instability (MSI-high, MSI-H) and microsatellite stable (microsatellitestability, MSS).

[0004] In 2017, the FDA approved the PD-1 inhibitor pembrolizumab (commonly known as drug K) for the treatment of solid tumors with microsatellite high instability (MSI-H) / mismatch repair deficiency (dMMR). Drug K is also known as the first anti-tumor immune drug that "does not look at the site but the marker". In the field of gynecological malignancies, the 2020 NCCN guidelines also pointed out that pembrolizumab is recommended for the second-line treatment of advanced endometrial cancer with MSI-H / dMMR and advanced cervical cancer with positive PD-L1 expression or MSI-H / dMMR.

[0005] At present, the commonly used method of MSI detection is multiplex fluorescence PCR combined with capillary electrophoresis. After detecting 5 microsatellite sites on the genome by labeling with different fluorescent groups, capillary electrophoresis is used to compare the difference in the length of microsatellite sequences between tumor tissues and normal tissues to determine whether there is MSI phenomenon; this method requires DNA from tumors and DNA from normal cells in the body, that is, they need to be paired, and PCR is performed on the two groups of samples for comparison analysis. In 2019, the "Chinese Expert Consensus on Microsatellite Instability Detection in Colorectal Cancer and Other Related Solid Tumors" pointed out that the combination of 5 to 7 marker sites based on the Pentaplex panel in the PCR method, BAT-25, BAT-26, NR-21, NR-22, NR-24, NR-27 and MONO-27, is recognized as the gold standard for MSIPCR detection; generally, instability of ≥2 sites is microsatellite high instability (MSI-H), instability of 1 site is microsatellite low instability (MSI-L), and no instability is microsatellite stable (MSS). The judgment criteria for the multiplex fluorescence PCR capillary electrophoresis spectrum can be subdivided into the following: (1) detecting whether there are new detection peaks in the tumor sample; (2) when there are less than 2 new detection peaks, detecting whether there are peaks in the tumor sample with a significantly increased peak height (more than 50%) relative to the product peak; (3) detecting whether there are 3 or more new continuous peaks in the tumor sample relative to the control sample (normal sample) (i.e., tailing phenomenon); (4) comparing the fragment size difference between the product peak of the tumor sample at a single site and the corresponding product peak in the control sample to see whether it is greater than 2 repeat units.

[0006] Due to the complexity of interpreting the results of multiplex fluorescence PCR capillary electrophoresis and the fact that the results mainly rely on the judgment of existing personnel, the results of multiplex fluorescence PCR capillary electrophoresis will be greatly affected by the subjective factors of the interpreters, and there is a certain probability of false negatives and false positives.

[0007] Therefore, there is an urgent need for a method that can reduce the subjective error in the interpretation of multiplex fluorescence PCR capillary electrophoresis results, so as to achieve a more accurate judgment of microsatellite instability. Summary of the invention

[0008] The purpose of the present invention is to overcome the above-mentioned shortcomings of the prior art and to provide a method for determining microsatellite instability based on fluorescent PCR capillary electrophoresis and its application.

[0009] The first object of the present invention is to provide a method for determining microsatellite instability based on fluorescence PCR capillary electrophoresis.

[0010] The second object of the present invention is to provide a model for determining microsatellite instability based on fluorescent PCR capillary electrophoresis.

[0011] The third object of the present invention is to provide a system for determining microsatellite instability.

[0012] A fourth object of the present invention is to provide a computer device.

[0013] A fifth object of the present invention is to provide a computer-readable storage medium.

[0014] In order to achieve the above object, the present invention is implemented by the following scheme:

[0015] A method for determining microsatellite instability based on fluorescent PCR capillary electrophoresis, characterized in that it comprises the following steps:

[0016] Data acquisition: respectively obtaining the fluorescence PCR capillary electrophoresis detection results of the test site and the normal site of the same microsatellite site; the normal site is a microsatellite stable site;

[0017] Data processing: Based on the fluorescence PCR capillary electrophoresis test results of the test site and the normal site obtained in the data acquisition step, the following processing is performed to obtain the matrix lateral displacement SD, the number of newly added peaks and the number of increased peaks with the minimum error score, specifically:

[0018] S1. Based on the fluorescence PCR capillary electrophoresis detection results of the test sites and the fluorescence PCR capillary electrophoresis detection results of the normal sites, a test site matrix and a normal site matrix of g rows × h columns are constructed respectively, with base numbers as rows and fluorescence peak values ​​as columns;

[0019] Each point in the matrix is ​​0 or 1, 0 means there is no fluorescence peak at that location, and 1 means there is a fluorescence peak at that location;

[0020] S2. The test site matrix and the normal site matrix obtained in step S1 are respectively intercepted, and according to the intercepted test site matrix and the normal site matrix, the intercepted test site matrix is ​​fixed and the error score is calculated based on the matrix lateral displacement to obtain the matrix lateral displacement SD when the error score is minimum;

[0021] Where SD≥0 means that the normal site matrix is ​​shifted to the right by SD columns after interception; SD<0 means that the normal site matrix is ​​shifted to the left by SD columns after interception;

[0022] S3. When the SD obtained in step S2 is ≥ 0, assign the 1 to (h-SD) columns of the normal site matrix obtained in step S1 to the new g-row × h-column matrix C (x,y) (1+SD)~h column, and C (x,y) Assign 0 to the 1~SD columns of the matrix C (x,y) , subtract the matrix C from the matrix of sites to be tested obtained in step S1 (x,y), obtain a site merging matrix of g rows × h columns, and assign values ​​of 1 to SD columns in the site merging matrix to 0;

[0023] When the SD obtained in step S2 is less than 0, the (1-SD)~h columns in the normal site matrix obtained in step S1 are assigned to the new g-row×h-column matrix C (x’,y’) 1~(h+SD) column, and C (x’,y’) The (h+1+SD)~h columns are assigned 0, and the matrix C is obtained (x’,y’) , subtract the matrix C from the matrix of sites to be tested obtained in step S1 (x’,y’) , obtain a site-merging matrix of g rows × h columns, and assign values ​​of (h+1+SD) to h columns in the site-merging matrix to 0;

[0024] S4. Set the noise baseline value and the newly added peak baseline value, first determine the newly added peak B by merging the site matrix obtained in step S3 (x’,y’) , and then determine the increased peak B in the remaining sites of the site merging matrix (x’',y”) ;

[0025] Among them, the newly added peak B (x’,y’) The point with a point value of 1 in the site merging matrix obtained in step S3, where 0<y'≤the number of columns in the site merging matrix, the noise baseline value≤x'≤the number of rows in the site merging matrix, and there is a point B in the column where the point is located (a,y’) , B (a,y’) When the value of is 0, a<the sum of the noise baseline value and the newly added peak baseline value, and 0<a<x'; where B (x’,y’) Represents the point value at row x' and column y' in the site merging matrix;

[0026] Increased peak B (x”,y”) The point C is the point with a midpoint value of 1 in the remaining sites of the site merging matrix, where 0<y”≤the number of columns of the site merging matrix, 0<x”≤the number of rows of the site merging matrix, and the point C in the column where the point is located has a point value of 0 (b,y”) , 0<b<x”, and when b takes the maximum value, x” minus b≥5; where B (x”,y”) Represents the point value at row x and column y in the site merging matrix;

[0027] S5. Perform quality control on the newly added peaks and the increased peaks in the site merge matrix determined in step S4, respectively, to obtain the newly added peak position information and the increased peak position information in the site merge matrix after quality control;

[0028] After quality control, the number of new peaks and the number of increased peaks in the site merging matrix are calculated by weighting according to the number of new peaks F and the number of increased peaks G in rows 1 to c, and the number of new peaks H and the number of increased peaks K in rows (c+1) to g;

[0029] Where c=g / 2 is rounded to an integer, the weight L of rows 1 to c in the site merging matrix ranges from [0, 0.4], and the weight M of rows (c+1) to g ranges from [0.6, 1];

[0030] Determination of microsatellite instability: The matrix lateral displacement SD, the number of newly added peaks and the number of increased peaks are input into the trained support vector machine, and the microsatellite instability of the tested site is determined according to the output results of the support vector machine.

[0031] Preferably, the sites to be measured in the data acquisition are one or more sites to be measured.

[0032] When the number of sites to be tested in data acquisition is multiple, the above method can realize the joint determination of microsatellite instability for the multiple sites to be tested.

[0033] More preferably, the detection sites in the data acquisition are one or more of BAT25, BAT6, D2S123, D5S346, D17S250 and PentaC.

[0034] Preferably, before constructing the matrix in step S1, the fluorescence PCR capillary electrophoresis detection results of the test site and the fluorescence PCR capillary electrophoresis detection results of the normal site are respectively converted into one-dimensional fluorescence peak data of the test site and one-dimensional fluorescence peak data of the normal site, and normalized respectively.

[0035] More preferably, the data are converted into one-dimensional fluorescence peak data using ABIF Manager.

[0036] More preferably, the normalization process is to perform normalization processes on the fluorescence peak value and the base number respectively;

[0037] The normalization of the fluorescence peak value is specifically as follows: normalizing the one-dimensional fluorescence peak value data of the tested site and the one-dimensional fluorescence peak value data of the normal site based on the maximum value and the minimum value of the fluorescence peak value respectively;

[0038] The normalization of the base number is specifically as follows: normalizing the one-dimensional fluorescence peak data of the site to be tested and the one-dimensional fluorescence peak data of the normal site based on the ratio respectively;

[0039] The ratio is the ratio of the difference between the base number distribution range of the test site or the normal site and the number of repeating unit bases of the test site or the normal site.

[0040] More preferably, when there are multiple sites to be tested, it is also necessary to eliminate leakage peaks in the one-dimensional fluorescence peak data of the sites to be tested and the one-dimensional fluorescence peak data of the normal sites;

[0041] The leakage peak in the one-dimensional fluorescence peak data of the test site is: in the fluorescence PCR capillary electrophoresis detection results of different test sites, the fluorescence peak with the smaller fluorescence peak value among the two fluorescence peaks with a fluorescence peak value difference of less than 10 times for the same test site;

[0042] The leakage peak in the one-dimensional fluorescence peak data of the normal site is: in the fluorescence PCR capillary electrophoresis detection results of different normal sites, the fluorescence peak with the smaller fluorescence peak value among the two fluorescence peaks with a difference of less than 10 times for the same normal site.

[0043] The method of the present invention converts the fluorescence PCR capillary electrophoresis detection results of the test site into one-dimensional fluorescence peak data, normalizes the one-dimensional peak data of the test site, performs data processing, and combines the trained support vector machine to accurately detect the microsatellite instability of the test site;

[0044] The method eliminates the differences of fluorescent PCR capillary electrophoresis of different sites to be tested through normalization processing, thereby realizing accurate detection of microsatellite instability of different sites by a trained support vector machine.

[0045] Preferably, in S1, g=200-480, h=150-400.

[0046] More preferably, g=288, h=216.

[0047] Preferably, the interception in step S2 is intercepting the data of the first 85% to 95% rows in the matrix.

[0048] Preferably, the fixing of the intercepted matrix of the positions to be measured and the lateral displacement based on the matrix in step S2 specifically includes: fixing the intercepted matrix of the positions to be measured and lateral displacement of the intercepted normal position matrix.

[0049] Preferably, the error score calculation in step S2 is performed using an MSE function.

[0050] More preferably, the MSE function is as shown in Formula 1,

[0051] Formula I:

[0052] T is the overlapping part of the matrix of the sites to be tested after interception and the matrix of the normal sites after interception; N is the overlapping part of the matrix of the normal sites after interception and the matrix of the sites to be tested after interception; i is the row where each point in the matrix is ​​located, and the value range is 1 to g; j is the column where each point in the matrix is ​​located, and the value range is 1 to h; n = i×j.

[0053] Preferably, in step S4, the noise baseline value is 0-200, and the newly added peak baseline value is 20-60.

[0054] More preferably, the noise baseline value is 40, and the newly added peak baseline value is 35.

[0055] Preferably, the quality control in step S5 is performed using an open operation.

[0056] More preferably, the structural element for opening operation on the newly added peaks is a newly added peak structural matrix with 3 to 50 rows×3 columns, and the structural element for opening operation on the increased peaks is a increased peak structural matrix with 3 to 30 rows×3 columns.

[0057] Further preferably, the structural element for performing an opening operation on the newly added peaks is a newly added peak structural matrix of 35 rows×3 columns, and the structural element for performing an opening operation on the increased peaks is a increased peak structural matrix of 5 rows×3 columns.

[0058] More preferably, the opening operation is corrosion and expansion; the opening operation formula is shown in Formula II;

[0059] Formula II:

[0060] Erosion: A is the site merging matrix; z is all sites in the site merging matrix; when the opening operation is performed on the newly added peaks, B is the newly added peak structure matrix; when the opening operation is performed on the increased peaks, B is the increased peak structure matrix.

[0061] Preferably, in step S5, the weight L of rows 1 to c of the site merging matrix is ​​0.2, and the weight M of rows (c+1) to g is 0.8.

[0062] Preferably, the weighted calculation of the number of newly added peaks and the number of increased peaks in the site merging matrix in step S5 is as follows:

[0063] The number of newly added peaks in the site merging matrix = F*L+H*M;

[0064] The number of increased peaks in the site-merging matrix = G*L+K*M.

[0065] Preferably, the trained support vector machine is a support vector machine trained using a training set according to k-fold cross validation; the training set contains positive data and negative data, the expected output of the positive data in the support vector machine is 1, and the expected output of the negative data in the support vector machine is 0;

[0066] Among them, the positive data are the matrix lateral displacement, the number of newly added peaks and the number of increased peaks when the minimum error score of the microsatellite unstable site is obtained by replacing the site to be tested in the above method with the microsatellite unstable site and performing the data acquisition step and the data processing step;

[0067] Negative data are the matrix lateral displacement, number of newly added peaks and number of increased peaks when the minimum error score of the microsatellite stable site is obtained by replacing the site to be tested in the above method with the microsatellite stable site and performing the data acquisition step and the data processing step.

[0068] The microsatellite stable sites in the negative data and the normal sites in the above method are non-repeated samples, that is, the microsatellite stable sites that are normal sites cannot be used as negative data.

[0069] More preferably, the microsatellite stable loci and microsatellite unstable loci are both PCR-PCR-enhanced by Promega's OncoMate TM MSIDx Analysis System Manual (TM543) Operation Manual Determine whether microsatellites are stable or unstable.

[0070] More preferably, the k-fold cross validation is a k-fold cross validation with k=5.

[0071] Preferably, in the microsatellite instability judgment step, the output result of the trained support vector machine is 1, and the site to be tested is a microsatellite unstable site; the output result of the trained support vector machine is 0, and the site to be tested is a microsatellite stable site.

[0072] The present invention also claims a model for judging microsatellite instability based on fluorescent PCR capillary electrophoresis, the model comprising a data acquisition module, a data processing module, a prediction module and an output module;

[0073] The data acquisition module performs the data acquisition steps shown above on the sites to be tested and the normal sites; the normal sites are microsatellite stable sites;

[0074] The data processing module performs the data processing steps as shown above;

[0075] The prediction module performs the microsatellite instability determination step as shown above;

[0076] The output module is used to output the microsatellite instability result of the site to be tested obtained by the prediction module.

[0077] The present invention also claims protection for a determination system for determining microsatellite instability, comprising an acquisition unit, a storage unit and a processing unit, wherein the acquisition unit is used to obtain fluorescent PCR capillary electrophoresis detection results of a site to be tested and a normal site; the normal site is a microsatellite stable site;

[0078] The storage unit stores program instructions that can be executed by the processing unit;

[0079] The processing unit includes any of the models described above;

[0080] When the program instructions are executed by the processing unit, the fluorescence PCR capillary electrophoresis detection results of the sample to be tested and the normal sample acquired by the acquisition unit are input into any of the above-mentioned models contained in the processing unit to obtain the microsatellite instability results of the sample to be tested.

[0081] The present invention also claims protection for a computer device, comprising a memory and a processor, wherein the memory stores a computer program executable on the processor;

[0082] When the computer program is executed by a processor, any of the above-mentioned models and / or the above-mentioned judgment systems are implemented.

[0083] The present invention also seeks to protect a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the above-mentioned model and / or the above-mentioned judgment system when executed by a processor.

[0084] Compared with the prior art, the present invention has the following beneficial effects:

[0085] The present invention provides a method for judging microsatellite instability based on fluorescent PCR capillary electrophoresis, wherein the method obtains the fluorescent PCR capillary electrophoresis detection result of the site to be tested and the fluorescent PCR capillary electrophoresis of the microsatellite stable site, and converts them into matrix data through data processing, thereby obtaining the matrix lateral displacement, the number of newly added peaks and the number of increased peaks at the minimum error score of the site to be tested; the matrix lateral displacement, the number of newly added peaks and the number of increased peaks at the minimum error score are input into a trained support vector machine to obtain the microsatellite instability result of the site to be tested. The method can automatically judge the microsatellite instability of the site to be tested based on the fluorescent PCR capillary electrophoresis of the site to be tested, eliminates the subjective error of the traditional judgment of microsatellite instability, and thus reduces the generation of false positives and false negatives. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Figure 1 This is a flow chart of the method for determining microsatellite instability based on fluorescent PCR capillary electrophoresis as shown in Example 4;

[0087] Figure 2 : This is a visualization result diagram of the DQN reinforcement learning algorithm in Example 5; a is a visualization result diagram of batch A samples, and b is a visualization result diagram of batch B samples. DETAILED DESCRIPTION

[0088] The present invention is further described in detail below in conjunction with the accompanying drawings and specific examples of the specification. The examples are only used to explain the present invention and are not used to limit the scope of the present invention. The test methods used in the following examples are conventional methods unless otherwise specified; the materials and reagents used are reagents and materials that can be obtained from commercial channels unless otherwise specified.

[0089] The samples, microsatellite loci and microsatellite unstable loci used in the examples of the present invention are all from Shenzhen Hypros Medical Laboratory. All samples and microsatellite loci information are authorized by the sample source before testing; the microsatellite loci and microsatellite unstable loci are based on Promega's OncoMate TM The MSIDx Analysis System Manual (TM543) operation manual determines whether it is a microsatellite stable site or a microsatellite unstable site.

[0090] Example 1 Data processing method for determining microsatellite instability based on fluorescent PCR capillary electrophoresis

[0091] 1. Data Acquisition

[0092] Use the microsatellite instability (MSI) detection kit (multiple fluorescence PCR capillary electrophoresis method) (National Medical Device Registration No. 20213400070) to perform fluorescence PCR capillary electrophoresis on each detection site of tumor tissue and each detection site of normal tissue according to its instructions, and obtain the fluorescence PCR capillary electrophoresis detection results of each detection site of tumor tissue (fsa file stored in ABIF format) and the fluorescence PCR capillary electrophoresis detection results of each detection site of normal tissue (fsa file stored in ABIF format); the normal tissue is a healthy tissue with stable microsatellites (all microsatellite sites are stable microsatellite sites);

[0093] The information of each detection site and its corresponding fluorescence channel is shown in Table 1.

[0094] Table 1 Information of each detection site and its corresponding fluorescence channel

[0095]

[0096] 2. Data Processing

[0097] For the fluorescence PCR capillary electrophoresis detection results (fsa files stored in ABIF format) of each detection site obtained from tumor tissues and normal tissues as shown in Table 1, the fluorescence PCR capillary electrophoresis of each detection site was processed according to the following method to obtain the matrix lateral displacement (SD), the number of newly added peaks NPS and the number of increased peaks EPS at the minimum error score of each detection site. The specific method is as follows:

[0098] 1. Acquisition of data to be analyzed

[0099] The fluorescence PCR capillary electrophoresis detection results of the detection sites of tumor tissue and normal tissue are converted into one-dimensional fluorescence peak data of the detection sites of tumor tissue and normal tissue respectively by using ABIFManager; then, the leakage peaks in the one-dimensional fluorescence peak data of tumor tissue and the leakage peaks in the one-dimensional fluorescence peak data of normal tissue are eliminated respectively, and the data to be analyzed of the detection sites of tumor tissue and the data to be analyzed of the detection sites of normal tissue are obtained;

[0100] The leakage peak in the one-dimensional fluorescence peak data of the tumor tissue is: in the fluorescence PCR capillary electrophoresis detection results of different detection sites of the tumor tissue, the fluorescence peak with the smaller fluorescence peak value among the two fluorescence peaks with a fluorescence peak value difference within 10 times at the same detection site;

[0101] The leakage peak in the one-dimensional fluorescence peak data of normal tissue is: in the fluorescence PCR capillary electrophoresis detection results of different detection sites of normal tissue, the fluorescence peak with a smaller fluorescence peak value among two fluorescence peaks with a fluorescence peak value difference within 10 times at the same detection site.

[0102] (II) Obtaining the matrix lateral displacement, number of newly added peaks and number of increased peaks when minimizing the error score

[0103] 1. Matrix construction

[0104] S11. Processing data:

[0105] According to the base number distribution range of each detection site in Table 1, the data to be analyzed at the tumor tissue detection site are intercepted respectively, and the data to be analyzed within the base number distribution range of the tumor tissue detection site is retained. The fluorescence peak is normalized according to the maximum and minimum values ​​of the fluorescence peak according to Formula III, and then the base number is normalized by the ratio of the difference in the base number distribution range of the detection site and the number of bases of the repeating unit of the detection site to obtain the normalized tumor tissue detection site matrix data.

[0106] Formula III: X norm =(XX min ) / (X max -X min ); where X norm is the normalized value of X, X is the fluorescence peak, X max is the maximum value of the fluorescence peak, X min is the minimum value of the fluorescence peak.

[0107] S12. Matrix construction:

[0108] The normalized tumor tissue detection site matrix data obtained in step S1 is scaled to construct a matrix of 288 rows × 216 columns, with the base numbers in the fluorescence PCR capillary electrophoresis detection results as rows and the fluorescence peaks in the fluorescence PCR capillary electrophoresis detection results as columns, to obtain a matrix a of the tumor tissue detection sites. (x,y) ;

[0109] Each point in the matrix is ​​0 or 1, 0 means there is no fluorescence peak at that location, and 1 means there is a fluorescence peak at that location; the upper left corner of the matrix of the tumor tissue detection site is a (1，1) , the lower right corner is a (288，216) .

[0110] According to the above method, the data to be analyzed of the normal tissue detection sites are used to construct a matrix b of normal tissue detection sites with 288 rows × 216 columns. (x,y) .

[0111] 2. Establishment of the merge matrix:

[0112] S21. Respectively intercept the matrices of rows 1 to 258 in the matrix of tumor tissue detection sites and the matrix of normal tissue detection sites obtained in step S12 to obtain a intercepted tumor tissue detection site matrix and a intercepted normal tissue detection site matrix.

[0113] S22. The intercepted tumor tissue detection site matrix obtained in step S21 is completely overlapped with the intercepted normal tissue detection site matrix, the intercepted tumor tissue detection site matrix obtained in step S21 is fixed, the intercepted normal tissue detection site matrix obtained in step S21 is subjected to matrix lateral displacement, and the MSE function is used to calculate based on the overlapping part of the intercepted tumor tissue detection site matrix and the intercepted normal tissue detection site matrix to obtain an error score, and the matrix lateral displacement (SD) at the minimum error score is recorded;

[0114] The step length of the matrix lateral displacement is 1 column. The matrix lateral displacement (SD) is a positive value, which means that the normal tissue detection site matrix is ​​translated to the right by SD columns after interception. The matrix lateral displacement (SD) is a negative value, which means that the normal tissue detection site matrix is ​​translated to the left by SD columns after interception.

[0115] The MSE function is shown in Formula I;

[0116] Formula I:

[0117] T is the overlapping part of the tumor tissue detection site matrix after interception and the normal tissue detection site matrix after interception; N is the overlapping part of the normal tissue detection site matrix after interception and the tumor tissue detection site matrix after interception; i is the row where each point in the matrix is ​​located, and the value range is 1 to 258; j is the column where each point in the matrix is ​​located, and the value range is 1 to 216; n = i×j.

[0118] S23. Matrix merging

[0119] (1) When SD obtained in step S22 is ≥ 0:

[0120] The matrix b of the normal tissue detection sites obtained in step S12 (x,y) The data from 1 to (216-SD) columns in the matrix are assigned to the new 288 rows × 216 columns matrix c. (x,y) (SD+1)~216 columns, and c (x,y) Assign values ​​of 1 to SD columns in , and get the matrix c (x,y) .

[0121] Subtract the value of each point in the matrix of tumor tissue detection sites obtained in step S12 from the matrix c (x,y) The values ​​of the points at the corresponding positions in the matrix are obtained to obtain the detection site merge matrix d (x,y) , and assign 0 to the 1 to SD columns in the detection sites;

[0122] Among them, the detection site merging matrix d (x,y) Each point in A (x,y) Indicates that the value range of x is 1 to 288, and the value range of y is 1 to 216; A (x,y) The value is -1, 0 or 1, 1 means that the fluorescence peak only exists in the matrix of each detection site of the tumor tissue, and 0 means that the fluorescence peak exists in or does not exist in the matrix of each detection site of the tumor tissue and the matrix c at the same time. (x,y) In the equation, -1 means that the fluorescence peak only exists in matrix c (x,y) middle.

[0123] (2) When SD obtained in step S22 is less than 0:

[0124] The matrix b of the normal tissue detection sites obtained in step S12 (x,y) The (1-SD) to 216 columns of data are assigned to the new 288-row × 216-column matrix c. (x,y) 1~(216+SD) columns, and c (x,y) The (217+SD)~216 columns in the matrix are assigned 0, and the matrix c is obtained. (x,y) .

[0125] Subtract the value of each point in the matrix of tumor tissue detection sites obtained in step S12 from the matrix c (x,y)The values ​​of the points at the corresponding positions in the matrix are obtained to obtain the detection site merge matrix d (x,y) , and assign the values ​​of (217+SD) to 216 columns in the detection sites to 0;

[0126] Among them, the detection site merging matrix d (x,y) Each point in A xy Indicates that the value range of x is 1 to 288, and the value range of y is 1 to 216; A (x,y) The value is -1, 0 or 1, 1 means that the fluorescence peak only exists in the matrix of each detection site of the tumor tissue, and 0 means that the fluorescence peak exists in or does not exist in the matrix of each detection site of the tumor tissue and the matrix c at the same time. (x,y) In the equation, -1 means that the fluorescence peak only exists in matrix c (x,y) middle.

[0127] 3. Obtaining the NPS of the newly added peak quantity and the EPS of the increased peak quantity

[0128] S31. Set the noise baseline value (YM) to 40 and the new peak baseline value (NPR) to 35.

[0129] First, the detection site merging matrix obtained in step S23 is used to determine the newly added peak B (x’,y’) :New peak B (x’,y’) Merge matrix A for detection sites (x,y) = 1, where 0<y'≤216, 40(YM)≤x'≤288, and there is a point B in the column where the point is located (a,y’) , B (a,y’) When the value of is 0, a<75(YM+NPR) and 0<a<x'; (x’,y’) Represents the point value at row x' and column y' in the site merging matrix;

[0130] Then determine the increase peak B in the remaining sites of the detection site merging matrix (x”,y”) : Increased peak B (x”,y”) A is the remaining site in the detection site merge matrix (x,y) = 1, where 0<y”≤216, 0<x”≤288, and the column where the point is located has a point C with a value of 0 (b,y”) , 0<b<x”, and when b takes the maximum value, x” minus b≥5; where B (x”,y”) Represents the point value at row x and column y in the site merging matrix;

[0131] Get the detection site merge matrix d (x,y) The newly added peak position information and the increased peak position information in ;

[0132] The newly added peak position information and the increased peak position information are specifically the rows and columns of the newly added peaks and the increased peaks in the detection site merge matrix;

[0133] S32. Merge matrix d based on detection sites (x,y) The newly added peak position information and the increased peak position information in the detection site are merged into the matrix d using the open operation (x,y) The newly added peaks and the increased peaks in the matrix are quality controlled respectively, and the newly added peak position information and the increased peak position information in the detection site merger matrix after quality control are obtained;

[0134] The newly added peak structure element when performing the opening operation on the newly added peak is a newly added peak structure matrix of 35 rows × 3 columns, and the increased peak structure element when performing the opening operation on the increased peak is an increased peak structure matrix of 5 rows × 3 columns. The opening operation formula is shown in Formula II;

[0135] Formula II:

[0136] Erosion: A is the detection site merging matrix d (x,y) (288 rows × 216 columns); z is the detection site merge matrix d (x,y) All sites in ; when the opening operation is performed on the newly added peaks, B is the newly added peak structure matrix (35 rows × 3 columns); when the opening operation is performed on the increased peaks, B is the increased peak structure matrix (5 rows × 3 columns);

[0137] S33. Based on the newly added peak position information and the increased peak position information in the post-quality control detection site merge matrix obtained in step S32, calculate the number of newly added peaks F and the number of increased peaks G located in rows 1 to 144 in the detection site merge matrix, and the number of newly added peaks H and the number of increased peaks K located in rows 145 to 288;

[0138] The weight L of the number of newly added peaks and the number of increased peaks in rows 1 to 144 is set to 0.2, and the weight M of the number of newly added peaks and the number of increased peaks in rows 145 to 288 is set to 0.8. The number of newly added peaks NPS and the number of increased peaks EPS of the detection site are calculated according to Formula IV and Formula V;

[0139] Formula IV: NPS = 0.2*F + 0.8*H;

[0140] Formula V: EPS = 0.2*G + 0.8*K.

[0141] Example 2 A machine learning model for judging microsatellite instability based on fluorescent PCR capillary electrophoresis

[0142] 1. Construction of machine learning model

[0143] 1. Obtaining training and test sets

[0144] The training set is a sample set containing 43 microsatellite unstable loci and 79 microsatellite stable loci;

[0145] The test set is a sample set containing 20 microsatellite unstable sites and 25 microsatellite stable sites;

[0146] Both microsatellite unstable sites and microsatellite stable sites include BAT25 site, BAT26 site, D2S123 site, D5S346 site, D17S250 site and PentaC site;

[0147] The microsatellite unstable sites in the training set and the test set were recorded as positive, and the microsatellite stable sites were recorded as negative.

[0148] Samples in which all microsatellite loci are microsatellite stable loci are regarded as normal tissues; the microsatellite stable loci contained in the training set and the test set and the microsatellite stable loci in normal tissues are non-repeated samples, that is, the microsatellite stable loci in normal tissues cannot be used as negative data in the training set or the test set.

[0149] 2. Data processing of training set and test set

[0150] A microsatellite instability (MSI) detection kit (multiple fluorescence PCR capillary electrophoresis method) (National Medical Device Registration No. 20213400070) was used to perform fluorescence PCR capillary electrophoresis on the microsatellite stable sites in normal tissues and each microsatellite unstable site and microsatellite stable site in the training set according to its instructions, and the fluorescence PCR capillary electrophoresis detection results of the microsatellite stable sites in normal tissues (fsa files stored in ABIF format) and the fluorescence PCR capillary electrophoresis detection results of each microsatellite unstable site and microsatellite stable site in the training set (fsa files stored in ABIF format) were obtained. According to the method of Example 1 (negative data is to replace the sites to be tested in the above method with microsatellite stable sites, and positive data is to replace the sites to be tested in the above method with microsatellite unstable sites), the matrix lateral displacement, the number of newly added peaks and the number of increased peaks at the minimum error score for each microsatellite unstable site and microsatellite stable site in the training set were obtained.

[0151] According to the method of Example 1 (negative data is to replace the test site in the above method with a microsatellite stable site, and positive data is to replace the test site in the above method with a microsatellite unstable site), the matrix lateral displacement, the number of newly added peaks, and the number of increased peaks when the minimum error score is obtained for each microsatellite unstable site and microsatellite stable site in the test set.

[0152] 3. Construction of machine learning model

[0153] The matrix lateral displacement, the number of newly added peaks and the number of increased peaks at the minimum error score of each microsatellite unstable site and microsatellite stable site in the training set are respectively used as input variables, and are input into a support vector machine (SVM) using a k-fold validation method (k=5). Training is performed based on the microsatellite instability results of each site in the training set output by the support vector machine (the expected output of the microsatellite unstable site is 1, and the expected output of the microsatellite stable site is 0), until the microsatellite instability results of each site in the training set output by the support vector machine are not significantly different from the actual microsatellite instability of each site in the training set, thus obtaining a trained support vector machine;

[0154] If the output result of the support vector machine is 1, the site is determined to be an unstable microsatellite site; if the output result of the support vector machine is 0, the site is determined to be a stable microsatellite site;

[0155] Then, the matrix lateral displacement, number of newly added peaks and number of increased peaks at the minimum error score of each microsatellite unstable site and microsatellite stable site in the test set were used as input variables, and the k-fold validation method (k=5) was used to input them into the trained support vector machine for cross-validation. The trained support vector machine had an F1 value ≥ 0.95, which was the trained support vector machine.

[0156] The output result of the trained support vector machine is 1, and the site is determined to be an unstable microsatellite site; the output result of the trained support vector machine is 0, and the site is determined to be a stable microsatellite site.

[0157] 2. Experimental Results

[0158] During the support vector machine training process, the results of training and testing the support vector machine using the training set and the test set are shown in Table 2.

[0159] Table 2 Support vector machine training and testing results

[0160] Average score Training set (122) Test set (45) Accuracy 0.983 0.98 Accuracy 0.977 0.983 Recall 0.977 0.964 F1 value 0.977 0.972

[0161] Accuracy: A=(TP+TN) / (TP+FN+FP+TN); Precision: P=TP / (TP+FP); Recall: R=TP / (TP+FN); F1 value: F1=2*P*R / (P+R); TP is true positive, FP is false positive, TN is true negative, and FN is false negative.

[0162] The results showed that when the trained support vector machine was cross-validated using the test set, the F1 value = 0.972 ≥ 0.95, and a well-trained support vector machine was obtained. The well-trained support vector machine can be used to accurately determine microsatellite unstable sites and microsatellite stable sites.

[0163] Example 3 Validation of a machine learning model for determining microsatellite instability based on fluorescent PCR capillary electrophoresis

[0164] 1. Experimental Methods

[0165] 1. Verification of microsatellite stability

[0166] A sample set containing 49 microsatellite unstable loci (positive) and 110 microsatellite stable loci (negative) was used as validation set 1; both microsatellite unstable loci and microsatellite stable loci included BAT25 loci, BAT26 loci, D2S123 loci, D5S346 loci, D17S250 loci and PentaC loci.

[0167] The microsatellite instability (MSI) detection kit (multiple fluorescence PCR capillary electrophoresis method) (National Medical Device Registration No. 20213400070) was used to perform fluorescence PCR capillary electrophoresis on the microsatellite stable sites in the normal tissue in Example 2 and each microsatellite unstable site and microsatellite stable site in the verification set 1 according to its instructions, and the fluorescence PCR capillary electrophoresis detection results of the microsatellite stable sites in the normal tissue (fsa files stored in ABIF format) and the fluorescence PCR capillary electrophoresis detection results of each microsatellite unstable site and microsatellite stable site in the verification set 1 (fsa files stored in ABIF format) were obtained. According to the method of Example 1 (each microsatellite unstable site and microsatellite stable site in the verification set 1 was used to replace the sites to be tested in the above method), the matrix lateral displacement, the number of new peaks and the number of increased peaks at the minimum error score of each microsatellite unstable site and microsatellite stable site in the verification set 1 were obtained.

[0168] The matrix lateral displacement, the number of new peaks and the number of increased peaks when minimizing MSE for each microsatellite unstable site and microsatellite stable site in the verification set 1 are used as input variables, respectively, and input into the trained support vector machine (SVM) obtained in Example 2. The trained support vector machine (SVM) outputs the microsatellite instability results of each microsatellite unstable site or microsatellite stable site in the verification set 1.

[0169] 2. Verification of microsatellite stability at each detection site

[0170] The 774 test loci (including 162 microsatellite unstable loci and 612 microsatellite stable loci, the types of test loci included BAT25 loci, BAT26 loci, D2S123 loci, D5S346 loci, D17S250 loci and PentaC loci) derived from 129 samples were randomly divided into validation set 2 containing 588 test loci and validation set 3 containing 186 test loci; the sample data of validation set 2 and validation set 3 are shown in Table 3.

[0171] Table 3 Sample data of validation set 2 and validation set 3

[0172]

[0173] A microsatellite instability (MSI) detection kit (multiple fluorescence PCR capillary electrophoresis method) (National Medical Device Registration No. 20213400070) was used to perform fluorescence PCR capillary electrophoresis on the microsatellite stable sites of the normal tissue in Example 2 and each site to be tested in validation set 2 and validation set 3 according to its instructions to obtain the fluorescence PCR capillary electrophoresis detection results of the microsatellite stable sites in the normal tissue (fsa files stored in ABIF format) and the fluorescence PCR capillary electrophoresis detection results of each site to be tested in validation set 2 and validation set 3 (fsa files stored in ABIF format). According to the method of Example 1 (using each microsatellite unstable site and microsatellite stable site in validation set 2 and validation set 3 to replace the sites to be tested in the above method, respectively), the matrix lateral displacement, the number of newly added peaks and the number of increased peaks at the minimum error score for each site to be tested in validation set 2 and validation set 3 were obtained.

[0174] The matrix lateral displacement, the number of newly added peaks and the number of increased peaks at the minimum error score of each site to be tested in verification set 2 and verification set 3 are respectively used as input variables, and are input into the trained support vector machine (SVM) obtained in Example 2. The trained support vector machine (SVM) outputs the microsatellite instability results of each site to be tested in verification set 2 and verification set 3, and calculates the accuracy, precision, recall and F1 value of different types of sites to be tested in verification set 2 and verification set 3 based on the actual microsatellite instability results of the sites to be tested.

[0175] 2. Experimental Results

[0176] 1. The results of testing using the trained support vector machine obtained in Example 2 combined with the verification set 1 are shown in Table 4.

[0177] Table 4 Detection results of validation set 1

[0178] Average score Validation set 1 (159 items) Accuracy 0.994 Accuracy 0.982 Recall 1 F1 value 0.99

[0179] The results show that when the trained support vector machine obtained in Example 2 is combined with the data processing method shown in Example 1 to judge the microsatellite instability of each site in the verification set 1, the F1 value in the judgment result is = 0.99≥0.95, indicating that the trained support vector machine obtained in Example 2 can be used to judge the microsatellite instability of the site.

[0180] 2. The results of using the trained support vector machine obtained in Example 2 in combination with the verification set 2 for detection are shown in Table 5, and the results of using the trained support vector machine obtained in Example 2 in combination with the verification set 3 for detection are shown in Table 6.

[0181] Table 5 Detection results of validation set 2

[0182]

[0183]

[0184] Table 6 Detection results of validation set 3

[0185]

[0186] The results show that when the trained support vector machine obtained in Example 2 is used in combination with the data processing method shown in Example 1 to judge the microsatellite instability of each test site in verification set 2 and verification set 3, the F1 value is greater than 0.95 for different test sites, and the average F1 value of verification set 2 is 0.987, and the average F1 value of verification set 3 is 0.989.

[0187] This shows that the trained support vector machine obtained in Example 2 combined with the data processing method shown in Example 1 can accurately determine the microsatellite instability of different sites.

[0188] Example 4 A method for determining microsatellite instability based on fluorescent PCR capillary electrophoresis

[0189] A method for determining microsatellite instability based on fluorescent PCR capillary electrophoresis, the flow chart is as follows Figure 1 As shown, the specific steps include:

[0190] 1. Data Acquisition

[0191] Tumor tissue is used as a sample to be tested, wherein the tumor tissue is colorectal cancer, gastric cancer or sebaceous gland cancer tissue; and samples in which all microsatellite loci are microsatellite stable loci are used as normal tissue;

[0192] A microsatellite instability (MSI) detection kit (multiple fluorescence PCR capillary electrophoresis method) (National Medical Device Registration No. 20213400070) was used to perform fluorescence PCR capillary electrophoresis on each detection site in the sample to be tested and normal tissue according to its instructions, and the fluorescence PCR capillary electrophoresis detection results of each detection site (BAT25 site, BAT26 site, D2S123 site, D5S346 site, D17S250 site and PentaC site) of the sample to be tested and normal tissue were obtained (fsa file stored in ABIF format), and processed according to the method shown in Example 1 to obtain the newly added peak position information and the increased peak position information in the merged matrix of the detection sites after quality control, and the matrix lateral displacement, the number of newly added peaks and the number of increased peaks at the minimum error score of each detection site were obtained.

[0193] 2. Visualization of the Difference Matrix

[0194] Using the fluorescence PCR capillary electrophoresis test results of the detection sites of normal tissue as the background, the plt function in Python was used to encode the newly added peak position information and the increased peak position information in the combined matrix of the detection sites after quality control obtained in step 1 into a picture (the horizontal direction of the picture is the number of bases, and the vertical direction is the fluorescence peak value), and the newly added peaks and increased peaks were marked in different colors in the picture, and the horizontal displacement (Shift) of the matrix at the minimum error score of the detection site, the number of newly added peaks (New peaks, NPS) of the detection site, the number of increased peaks (Extra peaks, EPS) of the detection site and the highest fluorescence peak (Maxpeaks) were marked in the upper left corner of the picture;

[0195] The newly added peaks are marked in black, and the increased peaks are marked in purple.

[0196] 3. Microsatellite instability judgment

[0197] The matrix lateral displacement, the number of newly added peaks and the number of increased peaks at the minimum error score of each detection site in the sample to be tested obtained in step 1 are respectively used as input variables, and are input into the trained support vector machine (SVM) obtained in Example 2. The trained support vector machine (SVM) outputs the microsatellite instability result of each detection site of the sample to be tested;

[0198] The result output by the trained support vector machine (SVM) is 1, and the detection site of the sample to be tested is an unstable microsatellite site; the result output by the trained support vector machine (SVM) is 0, and the detection site of the sample to be tested is a stable microsatellite site.

[0199] Based on the microsatellite instability results of each detection site of the test sample output by the trained support vector machine (SVM), the microsatellite instability of the test sample is judged, specifically: when there is at least one microsatellite unstable site in the test sample, the test sample is a microsatellite unstable test sample; when there is no microsatellite unstable site in the test sample, the test sample is a microsatellite stable test sample.

[0200] Example 5 A system for determining microsatellite instability

[0201] A system for determining microsatellite instability, comprising an acquisition unit, a storage unit and a processing unit;

[0202] The acquisition unit is used to obtain the fluorescence PCR capillary electrophoresis test results of the sample to be tested and the normal sample; wherein the sample to be tested is a tumor tissue, and the normal sample is a healthy tissue with stable microsatellites (all microsatellite loci are stable microsatellite loci);

[0203] The storage unit stores program instructions that can be executed by the processing unit;

[0204] The processing unit includes the machine learning model for determining microsatellite instability shown in Example 2;

[0205] When the program instructions are executed by the processing unit, the fluorescent PCR capillary electrophoresis detection results of the sample to be tested and the normal sample acquired by the acquisition unit are input into the machine learning model for judging microsatellite instability contained in the processing unit to obtain the microsatellite instability result of the sample to be tested.

[0206] Example 6 Optimization of processing data for determining microsatellite instability based on fluorescent PCR capillary electrophoresis

[0207] The four characteristic parameters in the data processing method shown in Example 1 are optimized, and the four characteristic parameters include the noise baseline (YM), the newly added peak baseline (NPR), the rows of the newly added peak structure matrix (NPM), and the columns of the enhanced peak structure matrix (EPM), as follows:

[0208] 1. Experimental Methods

[0209] The value range of the noise baseline value (YM) in the data processing method shown in Example 1 is set to 0-200; the value range of the newly added peak baseline value (NPR) is set to 20-60; the value range of the row of the newly added peak structure matrix (NPM) is 3-50; the value range of the row of the enhanced peak structure matrix (EPM) is 3-30; wherein the step size of the noise baseline (YM) is 5, and the step sizes of NPR, NPM and EPM are all 1.

[0210] The DQN reinforcement learning algorithm is used to optimize the four feature parameters in the four-dimensional feature parameter space. During the optimization process, the batch A samples and the batch B samples are respectively processed by the method shown in Example 1 in combination with different feature parameter combinations, and the trained support vector machine obtained in Example 2 is used to detect the microsatellite instability of the batch A samples and the batch B samples. The F1 values ​​of the batch A and batch B samples processed with different feature parameter combinations are used as indicators for visualization.

[0211] The total number of samples in batch A was 39, including 7 positive samples and 32 negative samples, and the total number of sites was 234 (39*6), including 16 positive sites;

[0212] The total number of samples in batch B is 30, including 16 positive samples and 14 negative samples, and the total number of sites is 180 (30*6), including 56 positive sites;

[0213] Among them, the positive samples are microsatellite unstable samples, the negative samples are microsatellite stable samples, the positive sites are microsatellite unstable sites, and the negative sites are microsatellite stable sites.

[0214] 2. Experimental Results

[0215] The visualization results of the DQN reinforcement learning algorithm are shown in the figure Figure 2 As shown, a is the visualization result diagram of batch A samples, and b is the visualization result diagram of batch B samples; the X-axis represents the newly added peak baseline value (NPR), the Y-axis represents the row of the enhanced peak structure matrix (EPM), the Z-axis represents the row of the newly added peak structure matrix (NPM), and the size of the point represents the noise baseline value (YM); the color of the point represents the size of the F1 value, and the brighter the color, the closer the F1 value is to 1.

[0216] The results show that for batch A samples, there are two areas in the four-dimensional feature parameter space where the best parameter combination can be obtained, namely the front lower area and the back upper area; while for batch B samples, the best parameter combination can be obtained in the front lower area of ​​the four-dimensional feature parameter space.

[0217] Results show that in the four-dimensional feature parameter space, the front lower area is the global optimal parameter, while the back upper area is the local optimal parameter for batch 1 samples. Therefore, the noise baseline value (YM) is determined to be 40, the newly added peak baseline value (NPR) is 35, the row of the newly added peak structure matrix (NPM) is 35, the newly added peak structure matrix is ​​a matrix of 35 rows × 3 columns, the row of the enhanced peak structure matrix (EPM) is 5, and the enhanced peak structure matrix is ​​a matrix of 5 rows × 3 columns.

[0218] 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 the protection scope of the present invention. For ordinary technicians in this field, other different forms of changes or modifications can be made based on the above descriptions and ideas. It is not necessary and impossible to list all the implementation methods here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A method for determining microsatellite instability based on fluorescent PCR capillary electrophoresis, characterized in that: The following steps are involved: Data acquisition: respectively obtain the fluorescence PCR capillary electrophoresis detection results of the test site and the normal site of the same microsatellite site; The normal locus is a microsatellite stable locus; Data processing: Based on the fluorescence PCR capillary electrophoresis test results of the test site and the normal site obtained in the data acquisition step, the following processing is performed to obtain the matrix lateral displacement SD, the number of newly added peaks and the number of increased peaks with the minimum error score, specifically: S1. Based on the fluorescence PCR capillary electrophoresis detection results of the test sites and the fluorescence PCR capillary electrophoresis detection results of the normal sites, a test site matrix and a normal site matrix of g rows × h columns are constructed respectively, with base numbers as rows and fluorescence peak values ​​as columns; Each point in the matrix is ​​0 or 1, 0 means there is no fluorescence peak at that location, and 1 means there is a fluorescence peak at that location; S2. The test site matrix and the normal site matrix obtained in step S1 are respectively intercepted, and according to the intercepted test site matrix and the normal site matrix, the intercepted test site matrix is ​​fixed and the error score is calculated based on the matrix lateral displacement to obtain the matrix lateral displacement SD when the error score is minimum; Where SD≥0 means that the normal site matrix is ​​shifted to the right by SD columns after interception; SD<0 means that the normal site matrix is ​​shifted to the left by SD columns after interception; S3. When the SD obtained in step S2 is ≥ 0, assign the 1 to (h-SD) columns of the normal site matrix obtained in step S1 to the new g-row × h-column matrix C (x,y) (1+SD)~h column, and C (x,y) Assign 0 to the 1~SD columns of the matrix C (x,y) , subtract the matrix C from the matrix of sites to be tested obtained in step S1 (x,y) , obtain a site merging matrix of g rows × h columns, and assign values ​​of 1 to SD columns in the site merging matrix to 0; When the SD obtained in step S2 is less than 0, the (1-SD)~h columns in the normal site matrix obtained in step S1 are assigned to the new g-row×h-column matrix C (x’,y’) 1~(h+SD) column, and C (x’,y’) The (h+1+SD)~h columns are assigned 0, and the matrix C is obtained (x’,y’) , subtract the matrix C from the matrix of sites to be tested obtained in step S1 (x’,y’) , obtain a site-merging matrix of g rows × h columns, and assign values ​​of (h+1+SD) to h columns in the site-merging matrix to 0; S4. Set the noise baseline value and the newly added peak baseline value, first determine the newly added peak B by merging the site matrix obtained in step S3 (x’,y’) , and then determine the increased peak B in the remaining sites of the site merging matrix (x’',y”) ; Among them, the newly added peak B (x’,y’) The point with a point value of 1 in the site merging matrix obtained in step S3, where 0<y'≤the number of columns in the site merging matrix, the noise baseline value≤x'≤the number of rows in the site merging matrix, and there is a point B in the column where the point is located (a,y’) , B (a,y’) When the value of is 0, a<the sum of the noise baseline value and the newly added peak baseline value, and 0<a<x'; where B (x’,y’) Represents the point value at row x' and column y' in the site merging matrix; Increased peak B (x ”, y”) is the point with a midpoint value of 1 in the remaining sites of the site merging matrix, where 0<y”≤the number of columns of the site merging matrix, 0<x”≤the number of rows of the site merging matrix, and the point C in the column where the point is located has a point value of 0 (b,y”) , 0<b<x”, and when b takes the maximum value, x” minus b≥5; where B (x ",y") represents the point value at row x" and column y in the site merging matrix; S5. Perform quality control on the newly added peaks and the increased peaks in the site merge matrix determined in step S4, respectively, to obtain the newly added peak position information and the increased peak position information in the site merge matrix after quality control; After quality control, the number of new peaks and the number of increased peaks in the site merging matrix are calculated by weighting according to the number of new peaks F and the number of increased peaks G in rows 1 to c, and the number of new peaks H and the number of increased peaks K in rows (c+1) to g; Where c=g / 2 is rounded to an integer, the weight L of rows 1 to c in the site merging matrix ranges from [0, 0.4], and the weight M of rows (c+1) to g ranges from [0.6, 1]; Determination of microsatellite instability: The matrix lateral displacement SD, the number of newly added peaks and the number of increased peaks are input into the trained support vector machine, and the microsatellite instability of the tested site is determined according to the output results of the support vector machine.

2. The method according to claim 1, characterized in that The sites to be measured in the data acquisition are one or more sites to be measured.

3. The method according to claim 2, characterized in that The sites to be tested are one or more of BAT25, BAT6, D2S123, D5S346, D17S250 and PentaC.

4. The method according to claim 1, characterized in that: Before constructing the matrix in step S1, the fluorescence PCR capillary electrophoresis detection results of the test site and the fluorescence PCR capillary electrophoresis detection results of the normal site are respectively converted into one-dimensional fluorescence peak data of the test site and one-dimensional fluorescence peak data of the normal site, and normalized respectively.

5. The method according to claim 4, characterized in that The normalization process is to perform normalization process on the fluorescence peak value and the base number respectively; The normalization of the fluorescence peak value is specifically as follows: normalizing the one-dimensional fluorescence peak value data of the tested site and the one-dimensional fluorescence peak value data of the normal site based on the maximum value and the minimum value of the fluorescence peak value respectively; The normalization of the base number is specifically as follows: normalizing the one-dimensional fluorescence peak data of the site to be tested and the one-dimensional fluorescence peak data of the normal site based on the ratio respectively; The ratio is the ratio of the difference between the base number distribution range of the test site or the normal site and the number of repeating unit bases of the test site or the normal site.

6. The method according to claim 1, characterized in that In step S1, g=200-480, h=150-400.

7. The method according to claim 1, characterized in that The error score calculation in step S2 is to calculate the error score using the MSE function.

8. The method according to claim 1, characterized in that The quality control in step S5 is performed by using an open operation.

9. The method according to claim 8, characterized in that The structural element for opening operation for newly added peaks is a newly added peak structural matrix of 3 to 50 rows×3 columns, and the structural element for opening operation for increased peaks is a increased peak structural matrix of 3 to 30 rows×3 columns.

10. A model for determining microsatellite instability based on fluorescent PCR capillary electrophoresis, characterized in that: The model includes a data acquisition module, a data processing module, a prediction module and an output module; The data acquisition module performs the data acquisition steps as shown in claim 1 on the sites to be tested and the normal sites; the normal sites are microsatellite stable sites; The data processing module performs the data processing steps as shown in claim 1; The prediction module performs the microsatellite instability determination step as shown in claim 1; The output module is used to output the microsatellite instability result of the site to be tested obtained by the prediction module.

Images