Method and device for predicting functional mutation of ERCC2 gene
Through whole-genome sequencing and alignment, and the ERCC2 functional mutation prediction model is used to predict, the problem of difficult to measure ERCC2 gene driver mutations in the prior art is solved, and the function of quickly identifying ERCC2 driver mutations is achieved, and the chemotherapy decision-making and treatment effect of bladder cancer is improved.
Patent Information
- Application Number
- CN202411738430.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-29
- Publication Date
- 2025-03-21
AI Technical Summary
The prior art lacks a prediction method for directly measuring driver mutations of ERCC2 gene, and it is difficult to distinguish whether ERCC2 mutations are driven mutations or passenger mutations. The functional verification experiment is difficult, time periods are long, and costly.
By obtaining tumor gene data, whole genome sequencing and comparison, the distribution characteristics of genomic somatic cells were obtained, and they were input into the ERCC2 functional mutation prediction model for prediction processing, and the prediction results of ERCC2 driver mutation were obtained.
It has achieved the identification of ERCC2-driven mutations in a short period of time, which helps to make decisions on adjuvant chemotherapy methods for bladder cancer and improves clinical treatment effects.
Smart Images
Figure CN119673276A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of bioinformatics computing technology. More specifically, the present application relates to a method and device for predicting functional mutations of the ERCC2 gene. Background Art
[0002] The efficacy of cisplatin in bladder cancer is closely related to the functional loss mediated by the driver mutation of ERCC2. Studies have shown that there is a significant correlation between mutations in the ERCC2 gene and the response to cisplatin in bladder cancer patients. The frequency of ERCC2 mutations in cisplatin responders reached 38%, while it was only 6% in cisplatin non-responders. Almost all clinically observed ERCC2 mutations indicate the loss of nucleotide excision repair function and are associated with increased sensitivity to cisplatin, suggesting that ERCC2 mutations may have stronger clinical significance in bladder cancer patients receiving chemotherapy.
[0003] The ERCC2 mutations found in clinical practice are distributed throughout the gene, and targeted sequencing methods cannot detect all regions of the genome. Even if a gene mutation is detected in the ERCC2 gene, it is difficult to distinguish whether the gene mutation is a driver mutation or a passenger mutation by only considering the mutation characteristics of a specific region. The experimental operation of functionally verifying whether the ERCC2 mutation is a driver mutation is difficult, time-consuming and expensive. Therefore, the prior art lacks a prediction (or detection) method for directly measuring the driver mutation of the ERCC2 gene. Summary of the invention
[0004] The purpose of the implementation of this application is to provide a method and device for predicting functional mutations of the ERCC2 gene, which can identify ERCC2 driver mutations, help make decisions on adjuvant chemotherapy methods for bladder cancer, and improve clinical treatment effects. The embodiments of this application are mainly implemented through the following technical solutions:
[0005] In a first aspect of the embodiments of the present application, a method for predicting a functional mutation of an ERCC2 gene is provided, comprising:
[0006] Obtain tumor genetic data;
[0007] Obtaining a first mutation result based on the tumor gene data;
[0008] Obtaining genomic somatic mutation distribution characteristics based on the first mutation result;
[0009] The genomic somatic mutation distribution characteristics are input into the ERCC2 functional mutation prediction model, and the ERCC2 functional mutation prediction model performs prediction processing on the genomic somatic mutation distribution characteristics to obtain the prediction result of ERCC2 driver mutation.
[0010] According to one embodiment of the present application, the step of obtaining a first mutation result based on the tumor gene data includes:
[0011] Performing whole genome sequencing on the tumor gene data to obtain a first DNA sequence;
[0012] The first DNA sequence is compared with the standard gene hg38 to obtain the first mutation result.
[0013] According to one embodiment of the present application, the step of obtaining the distribution characteristics of genomic somatic mutations based on the first mutation result includes:
[0014] Set up multiple characteristic regions of the whole genome;
[0015] Collecting feature data corresponding to each of the feature regions in the first mutation result;
[0016] Calculating the number of mutations of all mutations in each of the feature data;
[0017] Generate a mutation matrix based on all the mutation numbers and all the feature regions;
[0018] Performing data conversion processing on each element in the mutation matrix to eliminate scale differences between different elements and form a target matrix;
[0019] The genomic somatic mutation distribution characteristics are extracted from the target matrix.
[0020] According to one embodiment of the present application, the training step of the ERCC2 functional mutation prediction model includes:
[0021] Obtain a training data set and a true label set, wherein each training data in the training data set has a one-to-one correspondence with one of the true labels in the true label set;
[0022] Inputting target training data into a support vector machine model, wherein the support vector machine model performs prediction processing on the target training data to obtain a training result, wherein the target training data is any training data in the training data set;
[0023] Calculating a loss function based on the training results and a true label corresponding to the target training data;
[0024] The first model parameter of the support vector machine model is adjusted based on the loss function to form the ERCC2 functional mutation prediction model.
[0025] According to one embodiment of the present application, after the step of adjusting the first model parameter of the support vector machine model based on the loss function to form the ERCC2 functional mutation prediction model, the training step of the ERCC2 functional mutation prediction model further includes:
[0026] The ERCC2 functional mutation prediction model was evaluated using a leave-one-out cross-validation method to obtain an evaluation result;
[0027] The ERCC2 functional mutation prediction model is adjusted based on the evaluation results.
[0028] According to one embodiment of the present application, after the step of adjusting the ERCC2 functional mutation prediction model based on the evaluation result, the step of training the ERCC2 functional mutation prediction model further includes:
[0029] Generate a confusion matrix based on all training results and the true label set;
[0030] Calculating accuracy, recall, precision and / or F1 score based on the confusion matrix;
[0031] The performance of the ERCC2 functional mutation prediction model is evaluated based on the accuracy, the recall, the precision and / or the F1 score.
[0032] According to one embodiment of the present application, after the step of evaluating the performance of the ERCC2 functional mutation prediction model based on the accuracy, the recall, the precision and / or the F1 score, the training step of the ERCC2 functional mutation prediction model further includes:
[0033] A sensitivity analysis is performed on the ERCC2 functional mutation prediction model to adjust the second model parameter of the ERCC2 functional mutation prediction model.
[0034] A second aspect of the embodiments of the present application provides a device for predicting functional mutations of the ERCC2 gene, comprising:
[0035] A tumor gene data acquisition module, used to acquire tumor gene data;
[0036] A first mutation result obtaining module, used to obtain a first mutation result based on the tumor gene data;
[0037] A genomic somatic mutation distribution feature acquisition module, used to obtain a genomic somatic mutation distribution feature based on the first mutation result;
[0038] A prediction module is used to input the genome somatic mutation distribution characteristics into an ERCC2 functional mutation prediction model, and the ERCC2 functional mutation prediction model performs prediction processing on the genome somatic mutation distribution characteristics to obtain a prediction result of an ERCC2 driver mutation.
[0039] In a third aspect of an embodiment of the present application, a terminal device is provided, comprising: a processor and a memory, the memory being used to store a computer program, the processor being used to call and run the computer program stored in the memory, and executing the steps of the method for predicting a functional mutation of the ERCC2 gene provided in the first aspect of the embodiment of the present application.
[0040] In a fourth aspect of the embodiments of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium is used to store a computer program, wherein the computer program enables a computer to execute the steps of the method for predicting functional mutations of the ERCC2 gene provided in the first aspect of the embodiments of the present application.
[0041] The beneficial effects of the embodiments of the present application include:
[0042] The embodiment of the present application designs an ERCC2 functional mutation prediction model for predicting the prediction results of ERCC2 driver mutations. Specifically, the embodiment of the present application obtains a first mutation result based on tumor gene data; obtains genomic somatic mutation distribution characteristics based on the first mutation result; inputs the genomic somatic mutation distribution characteristics into the ERCC2 functional mutation prediction model, and outputs the prediction results of ERCC2 driver mutations. Therefore, compared with the prior art, the embodiment of the present application can identify ERCC2 driver mutations in a short time, which is helpful for decision-making on adjuvant chemotherapy methods for bladder cancer and improves clinical treatment effects. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the conventional technology, the drawings required for use in the embodiments or the conventional technology descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0044] Figure 1 A flowchart of the method for predicting functional mutations of the ERCC2 gene of the present application in some embodiments;
[0045] Figure 2 Reference map for the support vector machine model to classify the target training data into WT or ERCC2 mutant;
[0046] Figure 3 For Figure 2The associated confusion matrix shows the accuracy of the biased cross-validation predictions for the GE cohort.
[0047] Figure 4 Reference figure showing the accuracy of predictions for TCGA BLCA samples for the confusion matrix;
[0048] Figure 5 The reference graph of the probability of ERCC2 mutation predicted by the support vector machine model;
[0049] Figure 6 Reference figure showing the importance of feature regions as predictors in the SVM model for sensitivity analysis;
[0050] Figure 7 The trinucleotide context mutations were used as reference graphs for target training data for the support vector machine model;
[0051] Figure 8 A reference graph showing the accuracy of the support vector machine model trained by trinucleotide context mutations;
[0052] Fig. 9 A reference graph for the importance of trinucleotides as predictors in the model;
[0053] Fig.10 This is a schematic diagram of the principle of the device for predicting functional mutations of the ERCC2 gene of the present application in some embodiments;
[0054] Fig.11 This is a functional block diagram of the terminal device of the present application in some embodiments. DETAILED DESCRIPTION
[0055] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present application, so the present application is not limited by the specific embodiments disclosed below.
[0056] It should be noted that the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.
[0057] The terms "exemplary" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.
[0058] The terms "comprises," "includes," or any other variations thereof are intended to cover a non-exclusive inclusion, for example, a process, method, system, product, or apparatus that comprises a series of steps or elements is not necessarily limited to those steps or elements explicitly listed but may include other steps or elements not explicitly listed or inherent to such process, method, product, or apparatus.
[0059] Unless otherwise defined, all technical and scientific terms used in the specification of this application have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application. The term "and / or" used in the specification of this application includes any and all combinations of one or more related listed items.
[0060] Bladder cancer is one of the most common malignant tumors of the urinary system. Currently, the incidence of bladder cancer ranks 9th among malignant tumors worldwide, and the mortality rate ranks 13th among malignant tumors. Bladder cancer varies in region, race, and gender, and can occur in people of all ages. The occurrence and development of bladder cancer is a complex, multifactorial, and multi-step pathological process. Its specific pathogenesis has not yet been elucidated, and both internal genetic factors and external environmental factors play an important role.
[0061] Bladder cancer can be divided into non-muscle-invasive bladder cancer (NMIBC) and muscle-invasive bladder cancer (MIBC). Among them, the standard treatment for NMIBC is TURBT (Transurethral Resection of Bladder Tumor), and the bladder instillation treatment plan is determined according to the risk of recurrence. MIBC, squamous cell carcinoma, adenocarcinoma, urachal carcinoma and other comprehensive treatments mainly based on surgery are firstly radical total cystectomy, and some patients can choose partial cystectomy. Preoperative neoadjuvant chemotherapy is recommended for T2-4aN0M0 bladder urothelial carcinoma, and adjuvant chemotherapy or radiotherapy is decided after surgery based on pathological results. Metastatic bladder cancer is mainly treated with systemic chemotherapy, and palliative surgery and radiotherapy can be used to relieve symptoms.
[0062] Cisplatin is a commonly used anticancer drug and is widely used to treat various malignant tumors. Its chemical name is cis-dichloro-diamineplatin (II) salt, which belongs to the platinum compound. Cisplatin can inhibit the growth and spread of cancer by interfering with the replication and repair process of cancer cell DNA. Radical cystectomy is the standard treatment for MIBC patients with clinical stage T2-4aN0M0, but its 5-year overall survival rate is about 50%. In order to improve the treatment effect, cisplatin-based combined neoadjuvant chemotherapy has been widely used, and neoadjuvant immunotherapy with immune checkpoint inhibitors is under normal exploration. Neoadjuvant therapy has a significant impact on the overall survival of patients who reach ypT0 or at least ypT2. Multiple randomized trials and meta-analyses have shown that MIBC patients receiving cisplatin-based neoadjuvant chemotherapy can significantly improve the complete remission rate of tumors and prolong the overall survival of patients, reduce the risk of death by 10%-13%, increase the 5-year overall survival rate by 5%-8%, and increase the 5-year survival rate of T3 patients by up to 11%.
[0063] Cisplatin-based neoadjuvant chemotherapy combined with radical cystectomy is the standard treatment for muscle-invasive urothelial bladder cancer (MIBC). However, there is significant variability in the response to cisplatin, and given its significant toxicity, only a small number of eligible patients receive cisplatin-based treatment. Therefore, identifying the efficacy of cisplatin-based neoadjuvant chemotherapy is crucial for the widespread application of neoadjuvant chemotherapy and improving patient prognosis, which can significantly improve patient survival.
[0064] ERCC2 (Excision Repair Cross-Complementation Group 2) is a DNA helicase and a member of the nucleotide excision repair (NER) pathway, which can repair intrastrand crosslinks generated by genomic toxins such as ultraviolet irradiation and platinum chemotherapy. The prevalence of ERCC2 gene mutations in bladder cancer is 10% to 15%, and mutations are distributed throughout the ERCC2 gene sequence. It is a huge challenge to identify whether these mutations cause loss of function.
[0065] The efficacy of cisplatin in bladder cancer is closely related to the functional loss mediated by the driver mutation of ERCC2. Studies have shown that there is a significant correlation between the mutation of the ERCC2 gene and the patient's response to cisplatin. The frequency of ERCC2 mutations in cisplatin responders reached 38%, while it was only 6% in cisplatin non-responders. Almost all clinically observed ERCC2 mutations indicate the loss of nucleotide excision repair function and are associated with increased sensitivity to cisplatin, indicating that ERCC2 mutations may have stronger clinical significance in patients receiving chemotherapy.
[0066] Somatic missense mutations in ERCC2 are associated with improved responses to cisplatin chemotherapy. However, the effects of single ERCC2 variants on NER capacity and cisplatin sensitivity are unclear. If the function caused by ERCC2 gene mutations is to be verified, a common approach is to use CRISPR / Cas9 technology to introduce ERCC2 mutations in bladder cancer cell lines, implant them into the bladders of nude mice, and treat the mice with cisplatin. It will take at least two weeks of observation to determine whether the cisplatin-treated group shows significant drug sensitivity to ERCC2 mutations. This functional experiment is difficult to operate, takes a long time, and is expensive.
[0067] The ERCC2 gene region is included in most somatic cell targeted sequencing panels. However, the ERCC2 mutations found in the clinic are distributed throughout the gene, and the targeted sequencing method cannot detect all regions of the genome. Even if the ERCC2 gene is detected to have a gene mutation, it is difficult to distinguish whether the gene mutation is a driver mutation or a passenger mutation by only considering the mutation characteristics of a specific region. The experimental operation of functionally verifying whether the ERCC2 mutation is a driver mutation is difficult, time-consuming, and expensive. The prior art lacks a prediction (or detection) method for directly measuring the ERCC2 gene driver mutation, as well as a reliable and effective detection method for the effect on the cell NER ability. To this end, the present application embodiment proposes a method for predicting functional mutations of the ERCC2 gene to overcome the above-mentioned defects. The present application embodiment innovatively uses the whole genome somatic mutation distribution characteristics of tumor samples to predict their ERCC2 mutation status, and for the first time uses whole genome data to verify the pathogenicity of ERCC2 mutations in non-hotspot regions.
[0068] The specific implementation of the present application is further described below in conjunction with the accompanying drawings.
[0069] like Figure 1 FIG. 1 is a flowchart of a method for predicting functional mutations of the ERCC2 gene provided in the first aspect of the present application. Figure 1 In the method, the prediction method of the functional mutation of the ERCC2 gene includes:
[0070] S1. Obtain tumor gene data.
[0071] The tumor gene data includes multiple tumor cells and multiple normal cells.
[0072] The tumor gene data can be obtained from any bladder cancer patient, that is, the tumor gene data is bladder cancer gene data.
[0073] S2. Obtaining a first mutation result based on the tumor gene data.
[0074] Furthermore, the S2 step includes:
[0075] S21. Perform whole genome sequencing on the tumor gene data to obtain a first DNA sequence.
[0076] The present application embodiment uses high-throughput sequencing (HTS) to perform the above-mentioned whole genome sequencing processing operation. The high-throughput sequencing technology can quickly and accurately obtain mutation data within the whole genome, including comprehensive sequencing of coding exons, introns, UTR regions and chromatin states.
[0077] In other embodiments, targeted sequencing technology of ERCC2 may be used to replace the high-throughput sequencing technology.
[0078] The S21 step can provide a detailed genomic mutation map, providing a solid data foundation for subsequent analysis, thereby improving the sensitivity and specificity of mutation detection.
[0079] S22. Compare the first DNA sequence with the standard gene hg38 to obtain the first mutation result.
[0080] The standard gene hg38 is the hg38 reference genome, which is a reference version of the human genome provided and maintained by the Institute of Genomics.
[0081] The embodiment of the present application can use BWA (Burrows-Wheeler Aligner) software, Bowtie2 software or BLAST (Basic Local Alignment Search Tool, a search tool based on a local alignment algorithm) to implement the above alignment operation. It should be understood that the embodiment of the present application does not limit the specific method of the above alignment operation, and those skilled in the art can set it according to actual needs.
[0082] The first mutation result at least includes a site where the first DNA sequence is inconsistent with the standard gene hg38.
[0083] S3. Obtaining genomic somatic mutation distribution characteristics based on the first mutation result.
[0084] The genomic somatic mutation distribution characteristics can represent the overall mutation feature distribution of the genome and identify key features therein.
[0085] Furthermore, the S3 step includes:
[0086] S31. Set multiple characteristic regions of the whole genome.
[0087] The multiple characteristic regions may refer to those shown in Table 1.
[0088] Table 1
[0089]
[0090]
[0091] In Table 1, DHS regions are the abbreviation of DNase hypersensitive sites, and DNasehypersensitive sites refer to sites identified as DNase hypersensitive sites.
[0092] Reptiming is the abbreviation of Replication timing regions, and Replication timing regions refer to different DNA replication timing regions.
[0093] CBS regions are the abbreviation of CTCF binding sites, which refer to the CTCF factor binding site regions.
[0094] CBS flank regions are the abbreviation of Flank regions of CTCF binding sites. Flankregions of CTCF binding sites refer to the flanking regions surrounding the CBS regions.
[0095] Coding exon is the abbreviation of Coding exons regions, which refers to the known or main coding exon regions.
[0096] Intron is the abbreviation of Introns regions, which refers to the intron region of the genome.
[0097] 5UTR is the abbreviation of 5'UTR regions, which refers to the 5'UTR region.
[0098] 3UTR is the abbreviation of 3'UTR regions, which refers to the 3'UTR region.
[0099] Intergenic regions refer to the regions between coding genes on the genome.
[0100] Open chromatin is the abbreviation of Open chromatin regions, which refers to open chromatin regions.
[0101] Close chromatin is the abbreviation of Close chromatin regions, which refers to some areas in the chromatin region that are in a compact state.
[0102] S32: Collect feature data corresponding to each of the feature regions in the first mutation result.
[0103] S33, calculating the number of mutations of all mutations in each of the feature data.
[0104] The step S33 refers to counting the number of sites in each of the characteristic data where the first DNA sequence is inconsistent with the standard gene hg38 to obtain the number of mutations.
[0105] S32: Generate a mutation matrix based on all the mutation numbers and all the feature regions.
[0106] Further, the step S32 includes:
[0107] S321. Divide each mutation number by the length of the characteristic region corresponding to each mutation number to obtain a mutation density corresponding to each of the characteristic regions.
[0108] S322. Determine the dimension of the mutation matrix according to the total number of all the feature regions.
[0109] The rows of the mutation matrix represent different feature regions (that is, each row corresponds to a feature region), and the columns of the mutation matrix represent the mutation density and the number of mutations.
[0110] S323: Fill the mutation matrix based on the dimension, all the feature regions, all the mutation numbers, and all the mutation densities, so as to obtain the mutation matrix.
[0111] S33, performing data conversion processing on each element in the mutation matrix to eliminate the scale difference between different elements and form a target matrix.
[0112] The step S33 can be understood as performing a standardization process on the mutation matrix.
[0113] The standardization process may be performed using a Z-score standardization method (also known as standard deviation standardization) or a Min-Max standardization method (also known as zero-one standardization). Of course, in other implementations, the standardization process is not limited to the above examples, and may be determined by those skilled in the art according to actual needs.
[0114] S34. Extracting genomic somatic mutation distribution characteristics from the target matrix.
[0115] S4. Inputting the genomic somatic mutation distribution characteristics into the ERCC2 functional mutation prediction model, the ERCC2 functional mutation prediction model performs prediction processing on the genomic somatic mutation distribution characteristics to obtain the prediction result of ERCC2 driver mutation.
[0116] Through the above technical solution, the embodiment of the present application designs an ERCC2 functional mutation prediction model for predicting the prediction results of ERCC2 driver mutations. Compared with the prior art, the embodiment of the present application can identify ERCC2 driver mutations in a short time, which helps to make decisions on adjuvant chemotherapy methods for bladder cancer and improve clinical treatment effects.
[0117] The prediction results of the ERCC2 functional mutation prediction model classify the ERCC2 mutation status of the sample, which can improve the ability to identify the pathogenicity of the mutation and show excellent prediction performance, including high sensitivity, specificity and accuracy. In addition, the ERCC2 functional mutation prediction model can also reduce the situation where ERCC2 mutations are filtered out due to the low frequency of variant alleles, and can effectively predict the mutation function of unknown regions of ERCC2, providing potential biomarkers for clinical use, assisting in treatment decisions and prognosis assessments of bladder cancer. In some embodiments, the training steps of the ERCC2 functional mutation prediction model include:
[0118] S51, obtaining a training data set and a real label set, wherein each training data in the training data set has a one-to-one correspondence with one of the real labels in the real label set.
[0119] It should be understood that the data type and data format of each training data in the training data set are the same as the data type and data format of the genomic somatic mutation distribution feature.
[0120] The step of obtaining the training data set includes:
[0121] S511, obtaining multiple pieces of data to be processed;
[0122] The data to be processed are whole genome sequencing (WGS) data from 382 BLCA cohorts from Genomics England (GE). Among these 382 GE BLCA samples, 39 samples had ERCC2 hotspot mutations and were classified as ERCC2 mutations, and 343 samples with no protein-changing mutations in ERCC2 were classified as wild type (WT).
[0123] The multiple examples of data to be processed are all bladder cancer gene data.
[0124] S512, performing whole genome sequencing on the target data to be processed to obtain a second DNA sequence, wherein the target data to be processed is any one of the multiple data to be processed.
[0125] S513. Compare the second DNA sequence with the standard gene hg38 to obtain the second mutation result.
[0126] S514. Obtain training distribution features based on the second mutation result.
[0127] The data type and data format of the training distribution feature and the genomic somatic mutation distribution feature are the same.
[0128] S515: Combine all the training distribution features to form the training data set.
[0129] It should be understood that the embodiment of the present application uses the mutation density corresponding to the characteristic region of the whole genome to characterize the training distribution characteristics, and can also use the trinucleotide context mutation of the whole genome to characterize the training distribution characteristics.
[0130] S52: Input the target training data into a support vector machine model, and the support vector machine model performs prediction processing on the target training data to obtain a training result, wherein the target training data is any training data in the training data set.
[0131] The support vector machine model (SVM model) is capable of distinguishing driver mutations and passenger mutations in the target training data.
[0132] The support vector machine model is implemented by the svm function in the e1071 package of the R language.
[0133] Furthermore, in an embodiment of the present application, the training data set is split into a training set and a test set to train the support vector machine model.
[0134] Furthermore, the embodiments of the present application may also use other machine learning algorithms, such as gradient boosting machine (GBM), deep neural network (DNN) or convolutional neural network (CNN), to replace the support vector machine model with random forest or Lasso regression to train the ERCC2 functional mutation prediction model. In the process of model optimization, other machine learning algorithms, such as deep learning or AI models, may also be combined to further improve the prediction performance of the ERCC2 functional mutation prediction model.
[0135] Further, please refer to Figure 2 As shown, the support vector machine model classifies the target training data into WT ( Figure 2 Gray in the figure) or ERCC2 mutants ( Figure 2 The black part in the figure is a reference figure. Figure 2 The associated confusion matrix shows the accuracy of the biased cross-validation predictions for the GE cohort. Figure 3 See also Figure 4 , shows the classification of samples into WT or ERCC2 mutant from the described SVM model, and the associated confusion matrix shows the accuracy of the predictions for TCGA BLCA samples. Figure 4 In the TCGA, "TCGA-FD-A3N6", "TCGA-BL-A13J", "TCGA-DK-A1AE", "TCGA-BT-A20P", "TCGA-GD-A2C5", "TCGA-DK-A1AG", "TCGA-H4-A2HQ", "TCGA-DK-A1AA", "TCGA-BT-A20V", "TCGA-DK-A1A6", "TCGA-CF-A27C", "TCGA-BT-A20T", "TCGA-DK-A3IL", "TCGA-BT-A2LA", "TCGA-BT-A3PH", "TCGA-BT-A20Q", "TCGA-BT-A3PJ", "TCGA-FT-A3EE", "TCGA-FD-A3N5" and "TCGA-DK-A1A7" are all specific sample identifiers from the TCGA (The Cancer Genome Atlas) project and are used to identify and track specific cancer samples. You can also refer to Figure 5 , Figure 5 The figure shows the probability of ERCC2 mutation predicted by the support vector machine model for the 10 samples that were excluded from training.
[0136] S53: Calculate a loss function based on the training result and the true label corresponding to the target training data.
[0137] Specifically, the embodiment of the present application adopts the hinge loss function (Hinge Loss).
[0138] The calculation formula of the hinge loss function is:
[0139] L(y(w·x+b)=[1-y(w·x+b)];
[0140] Among them, y is the true label corresponding to the target training data (the value is ±1), w·x+b is the training result, w is the weight vector, x is the input feature vector (that is, the target training data), and b is the bias term.
[0141] S54. Adjusting the first model parameter of the support vector machine model based on the loss function to form the ERCC2 functional mutation prediction model.
[0142] The ERCC2 functional mutation prediction model can significantly improve the classification accuracy of driver mutations and passenger mutations, and the applicability of the model on different data sets has been verified, with good robustness.
[0143] In some embodiments, after the step S54, the step of training the ERCC2 functional mutation prediction model further comprises:
[0144] S55. Use the Leave-One-Out Cross-Validation (LOOCV) method to evaluate the ERCC2 functional mutation prediction model to obtain an evaluation result.
[0145] Furthermore, the embodiment of the present application may also use K-fold cross validation or Bayesian optimization to replace the leave-one-out cross validation method. The specific method to be used may be determined by those skilled in the art according to actual needs.
[0146] The embodiment of the present application adopts the leave-one-out cross-validation method to evaluate the model, which can ensure the stability and generalization ability of the model.
[0147] S56. Adjusting the ERCC2 functional mutation prediction model based on the evaluation results.
[0148] In some embodiments, after the step of S56, the step of training the ERCC2 functional mutation prediction model further comprises:
[0149] S57: Generate a confusion matrix based on all training results and the true label set.
[0150] In the embodiment of the present application, the confusion matrix is generated using the confusionMatrix function in the caret package in the R language.
[0151] The confusion matrix is a tool for evaluating the performance of a classification model, which can display the comparison between the predicted results (i.e., all training results) and the actual results (i.e., the true label set), and can also be used for binary classification problems.
[0152] S58. Calculate the accuracy, recall, precision and / or F1 score based on the confusion matrix.
[0153] The specific calculation process of step S58 can be implemented by existing technologies, so this application will not go into details here.
[0154] In other implementations, the confusion matrix may also be used to calculate other performance indicators, and the embodiments of the present application are not limited to the above four performance indicators.
[0155] S59. Evaluate the performance of the ERCC2 functional mutation prediction model based on the accuracy, the recall, the precision and / or the F1 score.
[0156] In some embodiments, after the step S59, the training step of the ERCC2 functional mutation prediction model further comprises:
[0157] S60. Performing a sensitivity analysis on the ERCC2 functional mutation prediction model to adjust a second model parameter of the ERCC2 functional mutation prediction model.
[0158] The implementation of the S60 step can determine the relative importance of different mutation features in predicting the ERCC2 mutation status. Exemplarily, the mutation / expected mutation density feature of the CTCF (CCCTC binding factor, transcription factor encoded by the CTCF gene) binding site region (CBS, CTCF binding sites, CTCF factor binding site) is the most important, followed by the APOBEC mutation / expected ratio observed in the CBS, which illustrates the importance of the global features of somatic mutations, and the use of whole gene data can more accurately identify the functional loss of ERCC2 mutations.
[0159] For example, the embodiment of the present application may use the Importance function to calculate the importance score of each variable, and visualize the importance score, that is, to plot the importance score. For details, please refer to Figure 6 As shown, Figure 6 Reference figure showing the importance of feature regions as predictors in the SVM model for sensitivity analysis.
[0160] refer to Figure 7 , the present embodiment can also use trinucleotide context mutation as the input of the model. When using trinucleotide context mutation as the input of the model, the accuracy of the model can refer to Figure 8 The importance of trinucleotides as predictors in the model can be found in Fig. 9 shown.
[0161] In other embodiments, the embodiment of the present application can also use the fit function to fit the linear model to optimize the ERCC2 functional mutation prediction model. It should be understood that in the training stage of the model, the ERCC2 functional mutation prediction model trained by using the mutation density corresponding to the characteristic region of the whole genome to characterize the training distribution characteristics has an accuracy of 98.69%; the ERCC2 functional mutation prediction model trained by using the trinucleotide context mutation of the whole genome to characterize the training distribution characteristics has an accuracy of 95.83%. Although the accuracy of the ERCC2 functional mutation prediction model trained by trinucleotide context mutations is low, it can also preliminarily predict the functional loss of ERCC2 mutations.
[0162] During the testing phase of the model, the ERCC2 functional mutation prediction model achieved 100% accuracy in classifying WT and mutant samples. The testing of the ERCC2 functional mutation prediction model not only ensured the reliability and effectiveness of the model in different scenarios, but also trained the trinucleotide model, expanding the scope of use of the model.
[0163] It should be understood that the prediction method of the functional mutation of the ERCC2 gene provided in the embodiments of the present application is not limited to predicting the mutation status of ERCC2 in bladder cancer (ie, driver mutation or passenger mutation), and those skilled in the art can train the model according to actual needs to predict other urothelial cancers.
[0164] In actual application, the ERCC2 functional mutation prediction model can be embedded in the clinical diagnosis system to achieve real-time analysis and decision support.
[0165] like Fig.10 , which is a principle block diagram of a prediction device for ERCC2 gene functional mutation provided by the second aspect of the present application. Fig.10 In the method, the prediction device 100 for the functional mutation of the ERCC2 gene comprises:
[0166] Tumor gene data acquisition module 101, used to acquire tumor gene data;
[0167] A first mutation result obtaining module 102, configured to obtain a first mutation result based on the tumor gene data;
[0168] A genome somatic mutation distribution feature obtaining module 103, used to obtain a genome somatic mutation distribution feature based on the first mutation result;
[0169] The prediction module 104 is used to input the genomic somatic mutation distribution characteristics into the ERCC2 functional mutation prediction model, and the ERCC2 functional mutation prediction model performs prediction processing on the genomic somatic mutation distribution characteristics to obtain the prediction result of ERCC2 driver mutation.
[0170] A third aspect of the embodiments of the present application provides a terminal device. The principle block diagram of the terminal device can be as follows: Fig.11 As shown. The terminal device includes a processor, a memory, a network interface, a display screen and a temperature sensor connected via a system bus. The processor is used to provide computing and control capabilities. The memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the terminal device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a method for predicting functional mutations of the ERCC2 gene is implemented. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the temperature sensor is pre-set inside the terminal device to detect the operating temperature of the internal device.
[0171] Those skilled in the art will understand that Fig.11 The principle block diagram shown in the figure is only a block diagram of a partial structure related to the scheme of the present invention, and does not constitute a limitation on the terminal device to which the scheme of the present invention is applied. The specific terminal device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0172] In some embodiments, an embodiment of the present application provides a terminal device, comprising a processor and a memory, the memory being used to store a computer program, the processor being used to call and run the computer program stored in the memory, and executing the steps of the method for predicting a functional mutation of the ERCC2 gene provided in the first aspect of the embodiment of the present application.
[0173] In a fourth aspect of the embodiments of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium is used to store a computer program, wherein the computer program enables a computer to execute the steps of the method for predicting functional mutations of the ERCC2 gene provided in the first aspect of the embodiments of the present application.
[0174] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0175] The technical features of the above embodiments can be combined without changing the basic principles of the present application. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0176] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the patent application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the scope of patent protection of the present application shall be subject to the attached claims.
Claims
1. A method for predicting functional mutations of the ERCC2 gene, characterized in that: include: Obtain tumor genetic data; Obtaining a first mutation result based on the tumor gene data; Obtaining genomic somatic mutation distribution characteristics based on the first mutation result; The genomic somatic mutation distribution characteristics are input into the ERCC2 functional mutation prediction model, and the ERCC2 functional mutation prediction model performs prediction processing on the genomic somatic mutation distribution characteristics to obtain the prediction result of ERCC2 driver mutation.
2. The method for predicting functional mutation of ERCC2 gene according to claim 1, characterized in that: The step of obtaining a first mutation result based on the tumor gene data comprises: Performing whole genome sequencing on the tumor gene data to obtain a first DNA sequence; The first DNA sequence is compared with the standard gene hg38 to obtain the first mutation result.
3. The method for predicting functional mutation of ERCC2 gene according to claim 1, characterized in that: The step of obtaining the distribution characteristics of genomic somatic mutations based on the first mutation result comprises: Set up multiple characteristic regions of the whole genome; Collecting feature data corresponding to each of the feature regions in the first mutation result; Calculating the number of mutations of all mutations in each of the feature data; Generate a mutation matrix based on all the mutation numbers and all the feature regions; Performing data conversion processing on each element in the mutation matrix to eliminate scale differences between different elements and form a target matrix; The genomic somatic mutation distribution characteristics are extracted from the target matrix.
4. The method for predicting functional mutation of ERCC2 gene according to claim 1, characterized in that: The training steps of the ERCC2 functional mutation prediction model include: Obtain a training data set and a true label set, wherein each training data in the training data set has a one-to-one correspondence with one of the true labels in the true label set; Inputting target training data into a support vector machine model, wherein the support vector machine model performs prediction processing on the target training data to obtain a training result, wherein the target training data is any training data in the training data set; Calculating a loss function based on the training results and a true label corresponding to the target training data; The first model parameter of the support vector machine model is adjusted based on the loss function to form the ERCC2 functional mutation prediction model.
5. The method for predicting functional mutation of ERCC2 gene according to claim 4, characterized in that: After the step of adjusting the first model parameter of the support vector machine model based on the loss function to form the ERCC2 functional mutation prediction model, the training step of the ERCC2 functional mutation prediction model further includes: The ERCC2 functional mutation prediction model was evaluated using a leave-one-out cross-validation method to obtain an evaluation result; The ERCC2 functional mutation prediction model is adjusted based on the evaluation results.
6. The method for predicting functional mutation of ERCC2 gene according to claim 5, characterized in that: After the step of adjusting the ERCC2 functional mutation prediction model based on the evaluation result, the step of training the ERCC2 functional mutation prediction model further includes: Generate a confusion matrix based on all training results and the true label set; Calculating accuracy, recall, precision and / or F1 score based on the confusion matrix; The performance of the ERCC2 functional mutation prediction model is evaluated based on the accuracy, the recall, the precision and / or the F1 score.
7. The method for predicting functional mutation of ERCC2 gene according to claim 6, characterized in that: After the step of evaluating the performance of the ERCC2 functional mutation prediction model based on the accuracy, the recall, the precision and / or the F1 score, the step of training the ERCC2 functional mutation prediction model further includes: A sensitivity analysis is performed on the ERCC2 functional mutation prediction model to adjust the second model parameter of the ERCC2 functional mutation prediction model.
8. A prediction device for ERCC2 gene functional mutation, characterized in that: include: A tumor gene data acquisition module, used to acquire tumor gene data; A first mutation result obtaining module, used to obtain a first mutation result based on the tumor gene data; A genomic somatic mutation distribution feature acquisition module, used to obtain a genomic somatic mutation distribution feature based on the first mutation result; A prediction module is used to input the genome somatic mutation distribution characteristics into an ERCC2 functional mutation prediction model, and the ERCC2 functional mutation prediction model performs prediction processing on the genome somatic mutation distribution characteristics to obtain a prediction result of an ERCC2 driver mutation.
9. A terminal device, characterized in that: include: A processor and a memory, the memory being used to store a computer program, the processor being used to call and run the computer program stored in the memory to execute the steps of the method for predicting a functional mutation of the ERCC2 gene as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that: Used to store a computer program, wherein the computer program enables a computer to execute the steps of the method for predicting a functional mutation of the ERCC2 gene according to any one of claims 1 to 7.