Training method and system of central nervous system tumor classification model and storage medium

Abstract

The invention relates to the field of computer systems based on specific calculation models, and provides a training method of a central nervous system tumor classification model in order to efficiently distinguish central nervous system tumor types by using multiple methylation chips, which comprises the following steps: acquiring original data based on scanning results of three methylation chips, original data formats are unified; combining all the original data; carrying out homogenization processing on the merged data; the homogenized data is converted into Methyl Set type data, and the Methyl Set type data is obtained; removing part of probe data to obtain a data matrix; training based on a training set sample in the data matrix to obtain a random forest classification model, and predicting a test set sample by using the random forest classification model to obtain an original score; and based on the original scores, training a corrected score classification model based on a multiple linear regression algorithm. According to the application, the efficiency of distinguishing the types of the central nervous system tumors by using multiple methylated chips is improved.

Description

Technical Field

[0001] The present application relates to the field of computer systems based on specific computing models, and in particular, to a method, system and storage medium for constructing a central nervous system tumor classification model. Background Art

[0002] DNA methylation is one of the phenomena of epigenetics. With the rise of precision medicine, DNA methylation plays a role in biological processes such as embryonic development, X chromosome inactivation, immune system regulation, and tumorigenesis. Methylation chip array technology is a high-throughput DNA methylation detection method developed based on microarray technology. It uses specific probes to bind to methylation sites on genomic DNA and analyzes the methylation status of different sites through fluorescent labeling.

[0003] Methylation chip: Methylation chip technology is a high-throughput DNA methylation detection method developed based on microarray technology. It uses specific probes to bind to methylation sites on genomic DNA and realizes methylation status analysis of different sites through fluorescent labeling. Currently, the Illumina methylation chip is commonly used. It is a widely used DNA methylation detection tool that can detect CpG sites on the whole genome with single-base resolution. The designed sites are mainly distributed in CpG islands, non-CpG methylation sites in stem cells, differential methylation sites of various tumors, promoters, enhancers and other regions. With the development of methylation chips, different versions have appeared iteratively. According to the number of CpG sites detected, they are divided into 450K (Infinium Human Methylation450), 850K (Methylation EPIC v1.0), and 935K (Methylation EPIC v2.0). Among them, the 850K (Methylation EPIC v1.0) chip was discontinued in 2023. The chip currently available on the market is the 935K (Methylation EPIC v2.0) chip.

[0004] Whole-genome methylation profile testing: The 2021 World Health Organization Classification of Central Nervous System Tumors (Fifth Edition) mentions: Currently, whole-genome methylation profile testing is an effective auxiliary method for the classification of brain and spinal cord tumors. It may be the most effective classification method for tumors with uncommon morphological characteristics, and it may be the only method to identify some rare tumors and subtypes. Methods for classifying central nervous system tumors based on whole-genome DNA methylation profiles have been reported in many papers. The current common tumor classification is based on the 2016 4th edition of the World Health Organization Classification of Central Nervous System Tumors, which divides 2,801 samples into 91 methylation categories (82 tumor types and 9 control sample types). Summary of the invention

[0005] The present application proposes a training method for a central nervous system tumor classification model, which is characterized by comprising: obtaining raw data based on the scanning results of a 450K methylation chip, an 850K methylation chip, and a 935K methylation chip; converting the format of the raw data of the 850K methylation chip and the 935K methylation chip into the format of the raw data of the 450K methylation chip; merging the converted data of the 850K methylation chip and the 935K methylation chip and the raw data of the 450K methylation chip to obtain merged data; based on the preprocessIllumina() function of the minfi package, The combined data are normalized to obtain normalized data; the normalized data are converted into MethylSet class data; the probe data of non-unique alignment, the probe data related to SNP and the probe data related to sex chromosome in the MethylSet class data are removed to obtain a data matrix; the data matrix is ​​split into a training set and a test set; a random forest classification model is trained based on the training set; the test set data is output as raw scores based on the forest classification model; based on the raw scores, a correction score classification model based on a multivariate linear regression algorithm is trained through the cv.glmnet() function of the glmnet package.

[0006] According to an implementation manner of the present application, obtaining raw data based on the scanning results of the 450K methylation chip, the 850K methylation chip, and the 935K methylation chip includes: converting the scanning results of the 450K methylation chip, the 850K methylation chip, and the 935K methylation chip into the raw data through the two functions read.metharray.sheet() and read.metharray.exp() of the minfi package, and the raw data is RGChannelSet class data.

[0007] According to the embodiment of the present application, obtaining raw data based on the scanning results of the 450K methylation chip, the 850K methylation chip, and the 935K methylation chip also includes:

[0008] The raw data converted from the 935K methylation chip were annotated using the annotation() function, and the annotation information was IlluminaHumanMethylationEPICv2.

[0009] According to an embodiment of the present application, converting the format of the raw data of the 850K methylation chip and the 935K methylation chip into the format of the raw data of the 450K methylation chip includes: converting the format of the raw data of the 850K methylation chip into the format of the raw data of the 450K methylation chip by using the combineArrays() function of the minfi package; converting the format of the raw data of the 935K methylation chip into the format of the raw data of the 850K methylation chip by using the convertArray_935k() function, and then converting the format of the raw data of the 850K methylation chip into the format of the raw data of the 450K methylation chip by using the combineArrays() function of the minfi package.

[0010] According to an embodiment of the present application, the steps performed by the convertArray_935k() function include: obtaining all probe numbers contained in the 935K methylation chip and the 850K methylation chip, taking the probes with the same probe numbers as the 850K methylation chip and the 935K methylation chip, and obtaining a first group of probes; dividing the first group of probes into Type I, Type II, Type SnpI, Type SnpII, and Type Control, and removing the probes with the same probe numbers in the 935K methylation chip and the 850K methylation chip but at least one of the color channel, probe sequence A, and probe sequence B being inconsistent according to each type, to obtain a second group of probes; removing the probe data that does not belong to the second group of probes in the original data of the 935K methylation chip, and modifying the annotation information of the original data of the 935K methylation chip to IlluminaHumanMethylationEPIC.

[0011] The present application also provides a training system for a central nervous system tumor classification model, the training system comprising: a memory storing executable instructions; and one or more processors communicating with the memory to execute the executable instructions to complete the following operations: acquiring raw data based on the scanning results of the 450K methylation chip, the 850K methylation chip and the 935K methylation chip; converting the format of the raw data of the 850K methylation chip and the 935K methylation chip into the format of the raw data of the 450K methylation chip; merging the converted data of the 850K methylation chip and the 935K methylation chip and the raw data of the 450K methylation chip to obtain merged data; The combined data are normalized using the preprocessIllumina() function of the minfi package to obtain normalized data; the normalized data are converted into MethylSet data; the probe data of non-unique alignment, the probe data related to SNP, and the probe data related to sex chromosomes in the MethylSet data are removed to obtain a data matrix; the data matrix is ​​split into a training set and a test set; a random forest classification model is trained based on the training set; the test set data is output as raw scores based on the forest classification model; based on the raw scores, a correction score classification model based on a multivariate linear regression algorithm is trained using the cv.glmnet() function of the glmnet package.

[0012] The present application also provides a computer-readable storage medium for training a central nervous system tumor classification model, wherein the computer-readable storage medium stores executable instructions, and the executable instructions can be executed by one or more processors to complete the following operations: obtaining raw data based on the scanning results of the 450K methylation chip, the 850K methylation chip, and the 935K methylation chip; converting the format of the raw data of the 850K methylation chip and the 935K methylation chip into the format of the raw data of the 450K methylation chip; merging the converted data of the 850K methylation chip and the 935K methylation chip and the raw data of the 450K methylation chip to obtain merged data; based on the p reprocessIllumina() function, perform homogenization on the merged data to obtain homogenized data; convert the homogenized data into MethylSet class data; remove non-uniquely aligned probe data, SNP-related probe data, and sex chromosome-related probe data in the MethylSet class data to obtain a data matrix; split the data matrix into a training set and a test set; train a random forest classification model based on the training set; output the test set data as raw scores based on the forest classification model; based on the raw scores, train a correction score classification model based on a multivariate linear regression algorithm through the cv.glmnet() function of the glmnet package.

[0013] The method for constructing a central nervous system tumor classification system provided in the present application uses multiple types of methylation chip data, achieves the compatibility of 450K methylation chip, 850K methylation chip and 935K methylation chip in one classification system, and improves the efficiency of using methylation chips to distinguish central nervous system tumor categories. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0015] Figure 1 is a schematic diagram of a training method for a central nervous system tumor classification model according to an embodiment of the present application;

[0016] Figure 2 is a sample information table according to an embodiment of the present application; and

[0017] Figure 3 It is a schematic block diagram of a training system for a central nervous system tumor classification model according to an embodiment of the present application. DETAILED DESCRIPTION

[0018] In order to better understand the present application, the technical solution of the present application will be described in more detail with reference to the accompanying drawings. It should be understood that these detailed descriptions are only descriptions of exemplary embodiments of the present application and are not intended to limit the scope of the present application in any way. Throughout the specification, the same reference numerals refer to the same elements. The expression "and / or" includes any combination or all combinations of one or more items in the associated listed items.

[0019] It should be noted that in this specification, expressions such as "first", "second", "third", etc. are only used to distinguish one feature from another feature, and do not represent any limitation on the features. Therefore, without departing from the teaching of this application, the first group of probes discussed below may also be referred to as the second group of probes. And vice versa.

[0020] In the accompanying drawings, the size, proportions and shapes of the figures have been slightly adjusted for ease of illustration. The drawings are for illustration only and are not drawn strictly to scale. As used herein, the terms "substantially," "approximately," and similar terms are used as terms of approximation, not degree, and are intended to account for the inherent deviations in measurements or calculations that would be recognized by one of ordinary skill in the art.

[0021] It should also be understood that expressions such as "including", "comprising", "having", "containing" and / or "comprising" are open rather than closed expressions in this specification, which indicate the presence of the stated features, elements and / or components, but do not exclude the presence of one or more other features, elements, components and / or combinations thereof. In addition, when expressions such as "at least one of..." appear after a list of listed features, they modify the entire list of features rather than just a single feature in the list. In addition, when describing embodiments of the present application, "may" is used to mean "one or more embodiments of the present application". And, the wording "exemplary" is intended to refer to an example or illustration.

[0022] Unless otherwise specified, all words (including engineering terms and scientific and technological terms) used in this article have the same meaning as those commonly understood by ordinary technicians in the field to which this application belongs. It should also be understood that, unless clearly stated in this application, words defined in common dictionaries should be interpreted as having the same meaning as their meaning in the context of the relevant technology, and should not be interpreted in an idealized or overly formal sense.

[0023] It should be noted that, in the absence of conflict, the features of the embodiments and examples in this application can be combined with each other. In addition, unless explicitly limited or contradictory to the context, the specific steps included in the method described in this application are not necessarily limited to the order described, but can be performed in any order or in parallel. The present application will be described in detail with reference to the accompanying drawings and in combination with the embodiments. Figure 1 Describe the training method of the classification model.

[0024] Figure 1 It is a flowchart of a training method for a central nervous system tumor classification model according to an embodiment of the present application.

[0025] In step S1010, raw data is obtained based on the scanning results of the 450K methylation chip, the 850K methylation chip, and the 935K methylation chip. The sample data information used for the training of the central nervous system tumor classification model mainly comes from public literature and hospital clinics. Among them, the raw data information of the samples in the public literature is mainly obtained through the 450K methylation chip or the 850K methylation chip; the raw data information of the hospital clinical samples is mainly obtained using the 850K methylation chip or the 935K methylation chip. Each sample is numbered and classified to form a sample information table.

[0026] See also Figure 2 The sample information table includes sample name, grouping information, chip number, chip location, tissue type, expected methylation classification and chip model. Among them, if the sample has no grouping information, the sample number is still filled in the grouping information of the sample information table. There are two types of tissue types in the sample information table, one is FFPE and the other is Frozen. Other tissue types will be treated as FFPE. Samples from different sources have different basis for expected methylation classification. Specifically, the data from public literature itself carries classification information, and its classification basis is 91 methylation classifications (82 tumor types and 9 control sample types) determined based on the results of unsupervised cluster analysis and the 4th edition of the World Health Organization classification of central nervous system tumors in 2016. The classification information from hospital clinical data is based on the results of histopathological examination, NGS analysis results and methylation chip classification results, with reference to the above 91 methylation classifications, and is determined by comprehensive judgment. The chip model refers to the 450K methylation chip, 850K methylation chip or 935K methylation chip.

[0027] According to one embodiment of the present application, in order to obtain the original data of hospital clinical samples, each sample is subjected to the steps of DNA extraction, sulfite conversion, DNA chip hybridization, staining, etc. according to the experimental process recommended by Illumina.

[0028] The idat data files of the methylation chips of different models are obtained by scanning and imaging. According to one embodiment of the present application, the idat data files of the methylation chips of different models are obtained by scanning using iScan or Nextseq 550.

[0029] According to the sample information table, the idat data files of each model chip are converted into raw data that can be processed by the model, that is, RGChannelSet class data. The content of RGChannelSet class data includes sample information, probe information, annotation information, optical signal values ​​of each probe, etc. Specifically, the idat data files obtained after scanning the 450K methylation chip or the 850K methylation chip are converted into RGChannelSet class data through the read.metharray.sheet() and read.metharray.exp() functions of the minfi package; the idat data files obtained after scanning the 935K methylation chip are converted into raw data RGChannelSet class data through the read.metharray.sheet() and read.metharray.exp() functions of the minfi package. Since there is no annotation information in the raw data of the 935K methylation chip, it is necessary to use the annotation() function to specify that the annotation information of the raw data of the 935K methylation chip is IlluminaHumanMethylationEPICv2. The annotation information IlluminaHumanMethylationEPICv2 is used to indicate that the methylation chip is a 935K methylation chip, so that the central nervous system tumor classification model can accurately identify the 935K methylation chip and perform special data processing on it during subsequent data processing.

[0030] In step S1020, the format of the original data of the 850K methylation chip and the 935K methylation chip is converted into the format of the original data of the 450K methylation chip; the converted data of the 850K methylation chip and the 935K methylation chip and the original data of the 450K methylation chip are merged to obtain merged data.

[0031] Different models of methylation chips contain different numbers and types of probes. Specifically, the 450K methylation chip contains 486,427 probes, the 850K methylation chip contains 866,552 probes, and the 935K methylation chip contains 937,690 probes. The probe types of the three methylation chips are cross-linked. The difference in the number and type of probes leads to inconsistent formats of raw data from different models of methylation chips, so it is necessary to unify the formats of raw data from the three methylation chips. Among them, the unification of the raw data formats of various models of methylation chips mainly refers to the screening and unification of information such as probe information, annotation information, and optical signal values ​​of each probe in the RGChannelSet class data of the raw data of various models of methylation chips.

[0032] According to an embodiment of the present application, the format of the raw data of the 850K methylation chip and the 935K methylation chip is converted into the format of the raw data of the 450K methylation chip. The conversion of the raw data of the 850K methylation chip and the 935K methylation chip into the format of the raw data of the 450K methylation chip requires different operations.

[0033] Specifically, for the raw data of the 935K methylation chip, the format of the 935K methylation chip raw data was first converted into the format of the 850K methylation chip raw data through the self-written convertArray_935k() function, and then the format of the 850K methylation chip raw data was converted into the format of the 450K methylation chip raw data through the combineArrays() function of the minfi package.

[0034] According to an embodiment of the present application, the specific process of convertArray_935k() function is as follows: obtain all probe numbers contained in the 935K methylation chip and the 850K methylation chip, take the probes with the same probe numbers in the 850K methylation chip and the 935K methylation chip, and obtain the first group of probes; divide the first group of probes into five groups according to Type I, Type II, Type SnpI, Type SnpII, and Type Control, and remove the probes with the same probe numbers in the 935K methylation chip and the 850K methylation chip in each group of probes, but at least one of the color channel, probe sequence A, and probe sequence B is inconsistent, to obtain the second group of probes; remove the probe data that does not belong to the second group of probes in the original data of the 935K methylation chip, and the obtained probe data is the completely consistent probe data contained in both the 935K methylation chip and the 850K methylation chip. Modify the annotation information of the original data of the 935K methylation chip from IlluminaHumanMethylationEPICv to IlluminaHumanMethylationEPIC. IlluminaHumanMethylationEPIC indicates that the format of the 935K methylation chip raw data has been converted to the format of the 850K methylation chip raw data.

[0035] For the original 850K methylation chip raw data and the 850K methylation chip raw data converted from the format of 935K methylation chip raw data, the combineArrays() function of the minfi package was used to convert the format of the 850K methylation chip format raw data into the format of the 450K methylation chip format raw data.

[0036] The raw data in the 450K methylation chip format converted from the raw data in the 935K methylation chip format and the raw data in the 450K methylation chip format converted from the raw data in the 850K methylation chip format were merged. Finally, the probe information in the merged data was the probe information contained in the raw data of the three types of methylation chips.

[0037] In step S1030, based on the preprocessIllumina() function of the minfi package, the merged data is homogenized to obtain homogenized data; the homogenized data is converted into MethylSet data. According to one embodiment of the present application, the merged data is homogenized by the preprocessIllumina() function of the minfi package, and the RGChannelSet data is converted into MethylSet data, that is, the M value of each probe.

[0038] In step S1040, probe data that are not uniquely aligned, probe data related to SNPs, and probe data related to sex chromosomes in the MethylSet class data are removed to obtain a data matrix. According to one embodiment of the present application, the data matrix consists of sample names and MethylSet class data corresponding to N probes in each sample. The sample names and the expected methylation classification corresponding to each sample are used to form a second table. When training and adjusting the random forest classification model, the second table and the data matrix are simultaneously input into the random forest classification model.

[0039] In step S1050, the data matrix is ​​split into a training set and a test set; a random forest classification model is trained based on the training set; the test set data is output as raw scores based on the random forest classification model; based on the raw scores, a correction score classification model based on a multivariate linear regression algorithm is trained through the cv.glmnet() function of the glmnet package.

[0040] According to one embodiment of the present application, the 10,000 most important probes are screened by the rfp() function of the randomForest package. The training set data consisting of the above 10,000 probes and their corresponding M values ​​is used to train a random forest classification model. The test set is input into the trained random forest classification model, and the random forest classification model performs classification prediction on the input data to obtain the original score of each sample under each classification type.

[0041] According to an embodiment of the present application, based on the raw scores of each sample in the test set under each classification type obtained by the random forest classification model, a correction score classification model based on a multiple linear regression algorithm is trained through the cv.glmnet() function of the glmnet package. The raw scores may appear to be very close, so that the classification model cannot accurately determine the classification result of each sample based on the raw scores of each sample under each classification type. The correction score classification model can be used to correct the raw scores and determine the classification result of the sample more clearly.

[0042] According to one embodiment of the present application, 12 FFPE samples from a hospital and confirmed by pathology were used for verification, of which 6 samples (samples 1-6) were tumor tissue samples and 6 were normal tissue control samples (samples 7-12). Sample DNA was extracted and central nervous system tumor classification analysis was performed using 850K methylation chip and 935K methylation chip respectively.

[0043] The samples were processed according to the experimental procedures recommended by Illumina, including DNA extraction, sulfite conversion, DNA chip hybridization, staining and other steps, and then the methylation chip was scanned and imaged using a scanner to obtain an idat data file. After data reading, data format conversion and merging, data homogenization, and probe filtering, the idat file was input into the random forest classification model for classification prediction to obtain the original score of each classification type. The original score was input into the correction score model obtained by training the multivariate linear regression algorithm to obtain the correction score of each classification type, and the top three subtypes of the correction score were output, and the classification type with the highest score was used as the final predicted classification result.

[0044] The test results are shown in the following table. After hybridization of the samples with the 850K methylation chip, the classification results of all samples were consistent with the pathological diagnosis results; after hybridization of the samples with the 935K methylation chip, the classification results of samples 1-3 and 5-12 were consistent with the pathological diagnosis results. Only the classification result of the tumor subtype of sample 4 was inconsistent with the pathological diagnosis result. For the 850K methylation chip, the accuracy of the methylation classification of the central nervous system tumor split model obtained after training was 100%; for the 935K methylation chip, the accuracy of the methylation classification of the central nervous system tumor split model obtained after training was 91.67%.

[0045]

[0046]

[0047] The present application also proposes a training system for a central nervous system tumor classification model, which can be implemented in the form of a mobile terminal, a personal computer (PC), a tablet computer, a server, etc. Figure 3 , which shows a structural schematic diagram of a system suitable for implementing the embodiments of the present application.

[0048] like Figure 3 As shown, the computer system includes one or more processors, a communication unit, etc. The one or more processors are, for example, one or more central processing units (CPUs) 301, and / or one or more graphics processing units (GPUs) 313, etc. The processors can perform various appropriate actions and processes according to executable instructions stored in a read-only memory (ROM) 302 or executable instructions loaded from a storage unit 308 to a random access memory (RAM) 303. The communication unit 312 may include, but is not limited to, a network card, and the network card may include, but is not limited to, an IB (Infiniband) network card.

[0049] The processor can communicate with the ROM 302 and / or the RAM 303 to execute the executable instructions, be connected to the communication unit 312 through the bus 304, and communicate with other target devices through the communication unit 312, so as to complete the operation corresponding to any one of the methods proposed in the embodiments of the present application, for example: obtaining raw data based on the scanning results of the 450K methylation chip, the 850K methylation chip and the 935K methylation chip; converting the format of the raw data of the 850K methylation chip and the 935K methylation chip into the format of the raw data of the 450K methylation chip; merging the converted data of the 850K methylation chip and the 935K methylation chip and the raw data of the 450K methylation chip to obtain the merged data; pre-processing based on the minfi package processIllumina() function, performs normalization processing on the merged data to obtain normalized data; converts the normalized data into MethylSet class data; removes non-uniquely aligned probe data, SNP-related probe data, and sex chromosome-related probe data in the MethylSet class data to obtain a data matrix; splits the data matrix into a training set and a test set; trains a random forest classification model based on the training set; outputs the test set data as raw scores based on the forest classification model; based on the raw scores, trains a correction score classification model based on a multivariate linear regression algorithm through the cv.glmnet() function of the glmnet package.

[0050] In addition, various programs and data required for the operation of the device may also be stored in RAM 303. CPU 301, ROM 302 and RAM 303 are connected to each other via bus 304. In the case of RAM 303, ROM 302 is an optional module. RAM 303 stores executable instructions, or writes executable instructions into ROM 302 during operation, and the executable instructions enable CPU 301 to perform operations corresponding to the above-mentioned communication method. Input / output interface (I / O interface) 305 is also connected to bus 304. Communication unit 312 can be integrated, or it can be configured to have multiple submodules (e.g., multiple IB network cards) and be linked on the bus.

[0051] The following components are connected to the I / O interface 305: an input section 306 including a keyboard, a mouse, etc.; an output section 307 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN card, a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 310 as needed.

[0052] If necessary, Figure 3 The architecture shown is only an optional implementation method. In the specific practice, the above can be modified according to actual needs. Figure 3 The number and type of components can be selected, deleted, added or replaced; different functional components can be set up by separate settings or integrated settings, for example, the GPU and the CPU can be set up separately or the GPU can be integrated on the CPU, the communication unit 312 can be set up separately or integrated on the CPU or GPU, etc. These replaceable implementations all fall within the protection scope disclosed in the present invention.

[0053] In particular, according to this application, the reference process Figure 1The described process can be implemented as a computer program product. For example, the present application proposes a computer program product, including computer-readable instructions, which implement the following operations when executed by a processor: obtaining raw data based on the scanning results of the 450K methylation chip, the 850K methylation chip, and the 935K methylation chip; converting the format of the raw data of the 850K methylation chip and the 935K methylation chip into the format of the raw data of the 450K methylation chip; merging the converted data of the 850K methylation chip and the 935K methylation chip and the raw data of the 450K methylation chip to obtain the merged data; preprocessingIllumina based on the minfi package a() function, normalize the merged data to obtain normalized data; convert the normalized data into MethylSet class data; remove non-uniquely aligned probe data, SNP-related probe data, and sex chromosome-related probe data in the MethylSet class data to obtain a data matrix; split the data matrix into a training set and a test set; train a random forest classification model based on the training set; output the test set data as raw scores based on the forest classification model; based on the raw scores, train a correction score classification model based on a multivariate linear regression algorithm through the cv.glmnet() function of the glmnet package.

[0054] In such an embodiment, the computer program product can be downloaded and installed from the network through the communication unit 309, and / or read and installed from the removable medium 311. When the computer program product is executed by the central processing unit (CPU) 301, the above functions defined in the method of the present application are performed.

[0055] The technical solution of the present application may be implemented in many ways. For example, the technical solution of the present application may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The sequence of steps used to illustrate the method is provided only for the purpose of more clearly illustrating the technical solution. Unless otherwise specified, the method steps of the present application are not limited to the sequence specifically described above. In addition, in some embodiments, the present application may also be implemented as a storage medium for storing a computer program product.

[0056] The above description is only an implementation method of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of protection involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the technical concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other.

Claims

1. A training method for a central nervous system tumor classification model, characterized in that: include: The raw data were obtained based on the scanning results of the 450K methylation chip, the 850K methylation chip, and the 935K methylation chip; Converting the format of the raw data of the 850K methylation chip and the 935K methylation chip into the format of the raw data of the 450K methylation chip; Merging the converted data of the 850K methylation chip and the 935K methylation chip and the original data of the 450K methylation chip to obtain merged data; Based on the preprocessIllumina() function of the minfi package, the merged data is homogenized to obtain homogenized data; Converting the homogenized data into MethylSet class data; Removing non-uniquely aligned probe data, SNP-related probe data, and sex chromosome-related probe data from the MethylSet class data to obtain a data matrix; Splitting the data matrix into a training set and a test set; Training a random forest classification model based on the training set; Outputting the test set data as raw scores based on the forest classification model; Based on the original scores, a correction score classification model based on a multiple linear regression algorithm is trained using the cv.glmnet() function of the glmnet package.

2. The classification method according to claim 1, characterized in that: The raw data obtained based on the scanning results of the 450K methylation chip, 850K methylation chip and 935K methylation chip include: The scanning results of the 450K methylation chip, the 850K methylation chip and the 935K methylation chip are converted into the raw data by using the read.metharray.sheet() and read.metharray.exp() functions of the minfi package. The raw data is RGChannelSet class data.

3. The classification method according to claim 1, characterized in that: The raw data obtained based on the scanning results of the 450K methylation chip, 850K methylation chip and 935K methylation chip also include: The raw data converted from the 935K methylation chip were annotated using the annotation() function, and the annotation information was IlluminaHumanMethylationEPICv2.

4. The classification method according to claim 1, characterized in that: Converting the format of the raw data of the 850K methylation chip and the 935K methylation chip into the format of the raw data of the 450K methylation chip includes: The format of the raw data of the 850K methylation chip is converted into the format of the raw data of the 450K methylation chip by using the combineArrays() function of the minfi package; The format of the raw data of the 935K methylation chip was converted into the format of the raw data of the 850K methylation chip by using the convertArray_935k() function, and then the format of the raw data of the 850K methylation chip was converted into the format of the raw data of the 450K methylation chip by using the combineArrays() function of the minfi package.

5. The classification method according to claim 3, characterized in that: The steps performed by the convertArray_935k() function include: Obtaining all probe numbers included in the 935K methylation chip and the 850K methylation chip, taking probes with the same probe numbers as those in the 850K methylation chip and the 935K methylation chip, to obtain a first group of probes; The first group of probes is divided into Type I, Type II, Type SnpI, Type SnpII, and Type Control, and probes having the same probe number but at least one inconsistent color channel, probe sequence A, and probe sequence B in the 935K methylation chip and the 850K methylation chip are removed in turn according to each type to obtain a second group of probes; The probe data that does not belong to the second group of probes in the original data of the 935K methylation chip are removed, and the annotation information of the original data of the 935K methylation chip is modified to IlluminaHumanMethylationEPIC.

6. A training system for a central nervous system tumor classification model, characterized in that: include: a memory storing executable instructions; as well as One or more processors, the one or more processors communicating with the memory to execute the executable instructions to perform the following operations: The raw data were obtained based on the scanning results of the 450K methylation chip, the 850K methylation chip, and the 935K methylation chip; Converting the format of the raw data of the 850K methylation chip and the 935K methylation chip into the format of the raw data of the 450K methylation chip; Merging the converted data of the 850K methylation chip and the 935K methylation chip and the original data of the 450K methylation chip to obtain merged data; Based on the preprocessIllumina() function of the minfi package, the merged data is homogenized to obtain homogenized data; Converting the homogenized data into MethylSet class data; Removing non-uniquely aligned probe data, SNP-related probe data, and sex chromosome-related probe data from the MethylSet class data to obtain a data matrix; Splitting the data matrix into a training set and a test set; Training a random forest classification model based on the training set; Outputting the test set data as raw scores based on the forest classification model; Based on the original scores, a correction score classification model based on a multiple linear regression algorithm is trained using the cv.glmnet() function of the glmnet package.

7. A computer-readable storage medium for training a central nervous system tumor classification model, characterized in that: The computer-readable storage medium stores executable instructions that can be executed by one or more processors to perform the following operations: The raw data were obtained based on the scanning results of the 450K methylation chip, the 850K methylation chip, and the 935K methylation chip; Converting the format of the raw data of the 850K methylation chip and the 935K methylation chip into the format of the raw data of the 450K methylation chip; Merging the converted data of the 850K methylation chip and the 935K methylation chip and the original data of the 450K methylation chip to obtain merged data; Based on the preprocessIllumina() function of the minfi package, the merged data is homogenized to obtain homogenized data; Converting the homogenized data into MethylSet class data; Removing non-uniquely aligned probe data, SNP-related probe data, and sex chromosome-related probe data from the MethylSet class data to obtain a data matrix; Splitting the data matrix into a training set and a test set; Training a random forest classification model based on the training set; Outputting the test set data as raw scores based on the forest classification model; Based on the original scores, a correction score classification model based on a multiple linear regression algorithm is trained using the cv.glmnet() function of the glmnet package.

Images