Method, equipment and program product for classifying glioma based on H3K27 mutation

Abstract

The invention relates to the field of intelligent medical treatment, in particular to an H3K27 mutation-based glioma classification method, equipment and a program product. Comprising the following steps: S1, acquiring mutation data of a to-be-detected sample, wherein the to-be-detected sample is a colloidal tumor sample containing H3K27 mutation; s2, classifying the mutation data to obtain a classification result of thalamus-like tumors or bridge-like tumors; wherein when the mutation characteristic is NF1 mutation, the tumor is judged to be a thalamic tumor; when the mutation characteristic is TP53 mutation, the tumor is judged to be a bridge brain-like tumor, and the novel subtype provided by the invention is beneficial to targeted therapy of the tumor and has a very good clinical value.

Description

Technical Field

[0001] The present application relates to the field of intelligent medical treatment, and specifically to a classification method, device, program product and computer-readable storage medium for gliomas based on H3 K27 mutation. Background Art

[0002] Diffuse midline gliomas (DMGs) with H3 K27 alterations that arise in the brain or spinal cord are among the most lethal tumors of the central nervous system, and effective treatments for these tumors are currently lacking. The main driver of these tumors is the global loss of H3 K27 trimethylation (H3 K27me3), which is mainly caused by oncohistones carrying H3 K27M mutations and rare H3 K27I mutations, as well as other rare events (such as EZHIP overexpression7-10). Mechanistically, the global loss of H3 K27me3 and the altered DNA methylation profile disrupt the normal development of neural progenitor cells, leading to their differentiation arrest. However, it is important to note that H3 K27 alterations alone are not sufficient to cause tumorigenesis, as at least one partner alteration (such as TP53 or PPM1D mutations, ACVR1 mutations, FGFR1 mutations, and others) is required for its neoplastic transformation. Studies have shown that different forms of H3 mutations and partner genetic alterations in brain DMGs are associated with different ages, anatomical locations, and clinical outcomes. Moreover, animal models have shown that H3 K27-altered tumors with different partners exhibit different growth rates, proliferation, and invasion phenotypes. Summary of the invention

[0003] In view of the problem that H3 K27-changed tumors with different partners show different growth rates, diffusion and invasion phenotypes, the present invention analyzes H3 K27-changed tumor-related omics data, proposes new subtypes of H3 K27-changed gliomas (thalamic-like tumors, pontine-like tumors) and a classification method for gliomas, specifically including: S1, obtaining mutation data of a sample to be tested, where the sample to be tested is a glial tumor sample including H3 K27 mutation; S2, classifying the mutation data to obtain a classification result of a thalamic-like tumor or a pontine-like tumor; wherein, when the mutation feature is NF1 mutation, it is determined to be a thalamic-like tumor; when the mutation feature is TP53 mutation, it is determined to be a pontine-like tumor.

[0004] The S2 also includes tumor location extraction, acquiring clinical data, and performing location feature extraction based on the clinical data to obtain location features. When the mutation feature is NF1 mutation and the tumor location is outside the pons, it is determined to be the thalamic-like tumor; when the mutation feature is TP53 mutation and the tumor location is inside the pons, it is determined to be the pons-like tumor.

[0005] Optionally, the classification is to cluster the mutation data to obtain a clustering result, and when the clustering feature includes NF1 mutation, it is determined to be a thalamic-like tumor, and when the clustering feature includes TP53 mutation, it is determined to be a pontine-like tumor.

[0006] The mutation classification is performed by feature extraction through a trained multi-omics integration model, and the training process of the multi-omics integration model is: obtaining a patient omics data set; constructing a patient similarity network set based on the omics data set; calculating the distance between similarity networks of different patients to obtain the weight of the edges between similarity networks; constructing a normalized weight matrix and a local affinity matrix of each similarity network based on the weight of the edges between the similarity networks; iteratively fusing the normalized weight matrix and the local affinity matrix to obtain a fused matrix; and clustering the fused matrix to obtain a clustering result.

[0007] Optionally, the training process further includes a normalization calculation, performing a normalization calculation on the fused matrix to obtain a fused normalized weight matrix, and clustering the fused normalized weight matrix to obtain a clustering result.

[0008] Optionally, the fused normalized matrix also includes a mean calculation, and the fused normalized matrix is ​​averaged to obtain a mean of the fused normalized matrix, and the mean of the fused normalized matrix is ​​clustered to obtain a clustering result.

[0009] Optionally, the clustering includes: a graph-based spectral clustering algorithm.

[0010] Optionally, the iterative fusion includes performing mean calculation on normalized weight matrices of different omics data types to obtain an average normalized matrix, and the average normalized matrix is ​​calculated with a local affinity matrix to obtain a fused matrix.

[0011] Optionally, the iterative fusion process is expressed as: Among them, v represents one of the different omics data, such as s represents somatic cells, c represents CNV, m represents methylation, r represents RNA-seq, A represents the local affinity matrix, P represents the normalized weight matrix, lab() is a function of the data type from different omics data, N represents the unique number of data types, T represents transposition, k represents the omics data label, k=1, Z(k) represents s; k=2, Z(k) represents c; k=3, Z(k) represents m; k=4, Z(k) represents r.

[0012] The omics data includes one or more of the following: somatic mutation data, copy number variation data, methylation data, and transcriptome data.

[0013] Optionally, S2 is replaced by S20: extracting mutation features from the tumor omics data to obtain mutation features; the method further includes S30: inputting the mutation features into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pons-like tumor.

[0014] Optionally, the classifier includes one or more of the following: convolutional neural network, Transformer, residual network, hole convolutional neural network, extreme learning machine, support vector machine, XGBoost, AdaBoost, random forest, naive Bayes.

[0015] The S2 is replaced by S21: extracting methylation features from the tumor omics data to obtain methylation features; the method also includes S31: inputting the methylation features into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pons-like tumor.

[0016] Optionally, the methylation feature is a MEX3A methylation site; Optionally, S31 is replaced by: classifying based on the methylation feature to obtain a classification result of a thalamic-like tumor or a pons-like tumor, wherein the methylation feature is a MEX3A methylation site, and when the MEX3A methylation site value is greater than or equal to a preset threshold, it is determined to be a thalamic-like tumor, and when the MEX3A methylation site value is less than the preset threshold, it is determined to be a pons-like tumor.

[0017] Optionally, S21 also includes position features, acquiring clinical data, extracting position features based on the clinical data to obtain position features, inputting methylation features and position features into a trained classifier for classification to obtain classification results of thalamic-like tumors or pontine-like tumors.

[0018] The S2 is replaced by S22: extracting gene features from the tumor omics data to obtain gene features; the S3 is replaced by S32: inputting the gene features into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pons-like tumor.

[0019] Optionally, the gene feature is MEX3A gene expression value.

[0020] Optionally, S22 also includes position features, acquiring clinical data, extracting position features based on the clinical data to obtain position features, inputting gene features and position features into a trained classifier for classification to obtain classification results of thalamic-like tumors or pontine-like tumors.

[0021] Optionally, S22 also includes methylation features, and methylation features are extracted from the tumor omics data to obtain methylation features, and the gene features and methylation features are input into a trained classifier for classification to obtain classification results of thalamic-like tumors or pontine-like tumors.

[0022] Optionally, the S22 also includes methylation features and position features; extracting methylation features from the tumor omics data to obtain methylation features; acquiring clinical data, and extracting position features based on the clinical data to obtain position features; inputting the gene features, methylation features, and position features into a trained classifier for classification to obtain classification results of thalamic-like tumors or pontine-like tumors.

[0023] Optionally, S2 is replaced by S23: acquiring tumor cell immune infiltration data, calculating the positive number of CD14 and / or CD8 in the cell immune infiltration data, and when the positive number of CD14 and / or CD8 is greater than a preset threshold, determining it to be a thalamic-like tumor; when the positive number of CD14 and / or CD8 is less than a preset threshold, determining it to be a pontine-like tumor; Optionally, the S20 also includes methylation features, gene features, position features, and immune infiltration features; extracting gene features from the tumor omics data to obtain gene features; obtaining clinical data, and extracting position features based on the clinical data to obtain position features; obtaining tumor cell immune infiltration data, and calculating the number of CD14 and / or CD8 positive cells in the cell immune infiltration data; inputting the mutation features, methylation features, gene features, position features, and immune infiltration features into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pontine-like tumor. The S20 also includes methylation features, which are extracted from the tumor omics data to obtain methylation features, and the mutation features and methylation features are input into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pontine-like tumor.

[0024] Optionally, the S20 also includes gene features, and gene features are extracted from the tumor omics data to obtain gene features, and the mutation features and gene features are input into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pons-like tumor.

[0025] Optionally, the S20 also includes methylation features and gene features, and gene features are extracted from the tumor omics data to obtain gene features, and the mutation features, methylation features, and gene features are input into a trained classifier for classification to obtain classification results of thalamic-like tumors or pons-like tumors.

[0026] Optionally, the S20 also includes methylation features, gene features, and position features; gene features are extracted from the tumor omics data to obtain gene features; clinical data are acquired, and position features are extracted based on the clinical data to obtain position features; the mutation features, methylation features, gene features, and position features are input into a trained classifier for classification to obtain classification results of thalamic-like tumors or pontine-like tumors.

[0027] Optionally, the S20 also includes immune infiltration features, obtaining tumor cell immune infiltration data, calculating the CD14 and / or CD8 positive counts in the cell immune infiltration data, and inputting the mutation features and immune infiltration features into a trained classifier for classification to obtain the classification results of thalamic-like tumors or pons-like tumors.

[0028] Optionally, the S20 also includes methylation features, gene features, and immune infiltration features. Gene features are extracted from the tumor omics data to obtain gene features, methylation features are extracted from the tumor omics data to obtain methylation features, tumor cell immune infiltration data are obtained, the CD14 and / or CD8 positive counts in the cell immune infiltration data are calculated, and the mutation features, methylation features, gene features, and immune infiltration features are input into a trained classifier for classification to obtain classification results of thalamic-like tumors or pons-like tumors.

[0029] The object of the present invention is to provide a computer program product, which includes a computer program or instructions, and the computer program or instructions are executed by a processor to implement the above-mentioned classification method of glioma based on H3 K27 mutation.

[0030] The object of the present invention is to provide a computer device, which includes a memory, a processor, and a computer program or instructions stored in the memory, wherein the computer program or instructions are executed by the processor to implement the above-mentioned classification method of glioma based on H3 K27 mutation.

[0031] The object of the present invention is to provide a computer-readable storage medium having a computer program or instruction stored thereon, wherein the computer program or instruction is executed by a processor to implement the above-mentioned classification method of glioma based on H3 K27 mutation.

[0032] Advantages of the present invention: 1. In response to the problem that H3 K27-altered tumors with different partners show different growth rates, diffusion and invasion phenotypes, the present invention proposes new glioma subtypes based on H3 K27 changes, including thalamic-like tumors and pontine-like tumors, and can provide tailored targeted treatments for different subtypes, thereby improving the effectiveness and reliability of treatments for different tumor subtypes and improving the treatment of tumor diseases.

[0033] 2. A classification method for the new subtype is proposed, and classification is performed based on mutation characteristics. When the extracted mutation characteristics are mainly NF1 mutations, it is a thalamic-like tumor; when the extracted mutation characteristics are mainly TP53 mutations, it is a pons-like tumor. This method can accurately distinguish the two new subtypes.

[0034] 3. Regarding the classification of new subtypes, the present invention also proposes tumor location characteristics. When the characteristics are NF1 mutation and the tumor location is outside the pons, it is a thalamic-like tumor; when the characteristics are TP53 mutation and the tumor location is inside the pons, it is a pontine-like tumor. This improves the accuracy of subtype classification and has practical clinical significance. It helps doctors to prioritize whether it is a thalamic-like tumor / pontine-like tumor when observing different tumor locations clinically, which helps to shorten the diagnosis process.

[0035] 4. For the classification of new subtypes, the present invention proposes classification features based on transcriptome and methylation group, including using MEX3A gene expression value and MEX3A methylation site as classification features for classification, and classifying new subtypes from different dimensions, which is helpful for mutual verification of classification results and improves the accuracy and reliability of classification. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0037] Figure 1 A schematic diagram of a flow chart of a method for classifying gliomas based on H3 K27 mutations provided in an embodiment of the present invention; Figure 2 A schematic diagram of a classification system for gliomas based on H3 K27 mutations provided in an embodiment of the present invention; Figure 3 A schematic diagram of a glioma classification device based on H3 K27 mutation provided in an embodiment of the present invention; Figure 4Figure 1 is a result of multi-omics clustering and subtype identification data analysis provided in the embodiments of the present invention. (B) Clinical information, data availability, and somatic mutations of H3-DSGs (H3-mutated diffuse spinal cord glioma) included in the present invention. C / T / L represent the cervical, thoracic, and lumbar segments of the human spinal cord, respectively. Summarize the diffuse midline glioma driver genes and reported key mutations identified by MutSigCV. (C) cSNF network showing two multi-omics subtypes of H3-DSGs. TP53 and NF1 mutation status are indicated by edge color. Edge width indicates the strength of evidence supporting patient-patient similarity from multi-omics data. The size of the nodes in the patient similarity network is positively correlated with the age of the patient at diagnosis (7 to 61 years old). The node color indicates the H3-DSG subtype defined by the cSNF method. (IV) Mutation frequency of key driver genes in different diseases. Brain-M-Pons and brain-M-Medulla are from Chen et al. (PMID: 32555164). (E) Diagnosis age of patients in different disease groups. (F) Kaplan-Meier survival curves of patients with tumors in different groups;

[0038] Figure 5 The constrained similarity network fusion typing method based on multi-omics data provided in the embodiment of the present invention, (A) a network-based multi-omics data integration analysis method. (B) a comparison of clinical pathological characteristics between subgroups using different grouping methods;

[0039] Figure 6 Functional annotation of differentially expressed genes between pontine-like and thalamic-like H3-DSGs provided in the embodiments of the present invention, (A) Volcano plot showing differentially expressed genes between two tumor subtypes based on bulk RNA-seq data. (B) Comparison of tumor purity of the two subtypes calculated based on WES data. (C and D) Comparison of lymphocyte infiltration indexes and myeloid infiltration indexes between pontine-like and thalamic-like subgroups using RNA-seq data;

[0040] Figure 7 (AB) Meta-network analysis provided for the embodiments of the present invention shows functions enriched in the pontine-like subtype (A) and thalamic-like subtype (B), respectively; (C) The proportion of malignant and non-malignant cells in each tumor, determined by single-cell RNA sequencing; Figure 8 The multiprotein staining provided in the embodiments of the present invention showed that H3K27M, CD31, MPO, CD14, CD3e, HLA-A, HLA-DR, and CD44 were expressed in both thalamic-like and pontine-like H3-DSGs tumors; Fig. 9(C) The proportion of cytotoxic CD8 T cells in pontine-like and thalamic-like H3-DSG provided in the examples of the present invention; (D) The expression of CD3e and GZMB proteins in pontine-like and thalamic-like subtype tumors; Scale bar: 50 μm; Fig.10 The ROC curve based on the classification of MEX3A methylation sites provided in the embodiment of the present invention; Fig.11 The ROC curve based on the MEX3A gene expression value classification provided in the embodiment of the present invention; Fig.12 A box plot of MEX3A methylation sites provided in an embodiment of the present invention; Fig.13 A box plot of the MEX3A gene expression values ​​provided in the embodiments of the present invention; Fig.14 Immunohistochemistry (IHC) provided for the examples of the present invention showed differences in thalamic-like and pontine-like CD14. DETAILED DESCRIPTION

[0041] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0042] In some of the processes described in the specification and claims of the present invention and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or executed in parallel. The sequence numbers of the operations, such as S101, S102, etc., are only used to distinguish between different operations, and the sequence numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit the "first" and "second" to be different types.

[0043] Figure 1 A schematic diagram of a glioma classification method based on H3 K27 mutation provided in an embodiment of the present invention specifically includes: S1: Obtain mutation data of a sample to be tested, where the sample to be tested is a glial tumor sample including H3 K27 mutation; In one embodiment, the mutation data is obtained through a gene testing kit.

[0044] In one embodiment, the mutation data is obtained by sequencing omics data.

[0045] S2: Classify the mutation data to obtain a classification result of a thalamic-like tumor or a pontine-like tumor; wherein, when the mutation feature is an NF1 mutation, it is determined to be a thalamic-like tumor; and when the mutation feature is a TP53 mutation, it is determined to be a pontine-like tumor.

[0046] In one embodiment, S2 also includes tumor location extraction, acquiring clinical data, and performing location feature extraction based on the clinical data to obtain location features. When the mutation feature is NF1 mutation and the tumor location is outside the pons, it is determined to be a thalamic-like tumor; when the mutation feature is TP53 mutation and the tumor location is inside the pons, it is determined to be a pons-like tumor.

[0047] In one embodiment, the classification is to cluster the mutation data to obtain a clustering result. When the clustering feature includes NF1 mutation, it is determined to be a thalamic-like tumor. When the clustering feature includes TP53 mutation, it is determined to be a pontine-like tumor.

[0048] In one embodiment, the mutation classification is performed by a trained multi-omics integration model, and the training process of the multi-omics integration model is: obtaining a patient omics data set; constructing a patient similarity network set based on the omics data set; calculating the distance between similarity networks of different patients to obtain the weight of the edges between similarity networks; constructing a normalized weight matrix and a local affinity matrix of each similarity network based on the weight of the edges between the similarity networks; iteratively fusing the normalized weight matrix and the local affinity matrix to obtain a fused matrix; and clustering the fused matrix to obtain a clustering result.

[0049] In one embodiment, the training process further includes a normalization calculation, performing a normalization calculation on the fused matrix to obtain a fused normalized weight matrix, and clustering the fused normalized weight matrix to obtain a clustering result.

[0050] Optionally, the fused normalized matrix also includes a mean calculation, and the fused normalized matrix is ​​averaged to obtain a mean of the fused normalized matrix, and the mean of the fused normalized matrix is ​​clustered to obtain a clustering result.

[0051] In one embodiment, the clustering includes: a graph-based spectral clustering algorithm.

[0052] In one embodiment, the iterative fusion includes performing mean calculation on normalized weight matrices of different omics data types to obtain an average normalized matrix, and the average normalized matrix is ​​calculated with a local affinity matrix to obtain a fused matrix.

[0053] In one embodiment, the iterative fusion process is expressed as: Among them, v represents one of the different omics data, such as s represents somatic cells, c represents CNV, m represents methylation, r represents RNA-seq, A represents the local affinity matrix, P represents the normalized weight matrix, lab() is a function of the data type from different omics data, N represents the unique number of data types, T represents transposition, k represents the omics data label, k=1, Z(k) represents s; k=2, Z(k) represents c; k=3, Z(k) represents m; k=4, Z(k) represents r.

[0054] In one embodiment, the omics data includes one or more of the following: somatic mutation data, copy number variation data, methylation data, and transcriptome data.

[0055] In one embodiment, S2 is replaced by S20: extracting mutation features from the tumor omics data to obtain mutation features; the method further includes S30: inputting the mutation features into a trained classifier for classification to obtain classification results of thalamic-like tumors or pons-like tumors.

[0056] In one embodiment, the classifier includes one or more of the following: convolutional neural network, Transformer, residual network, hole convolutional neural network, extreme learning machine, support vector machine, XGBoost, AdaBoost, random forest, naive Bayes.

[0057] In one embodiment, S2 is replaced by S21: extracting methylation features from the tumor omics data to obtain methylation features; the method further includes S31: inputting the methylation features into a trained classifier for classification to obtain classification results of thalamic-like tumors or pons-like tumors.

[0058] In one embodiment, the methylation feature is a MEX3A methylation site.

[0059] In one embodiment, S31 is replaced by: classifying based on the methylation feature to obtain a classification result of a thalamic-like tumor or a pons-like tumor, wherein the methylation feature is a MEX3A methylation site, and when the MEX3A methylation site value is greater than or equal to a preset threshold, it is determined to be a thalamic-like tumor, and when the MEX3A methylation site value is less than the preset threshold, it is determined to be a pons-like tumor.

[0060] In a specific embodiment, the preset threshold for classifying the MEX3A methylation sites is used to perform statistical analysis on the beta values ​​of the MEX3A methylation sites in the sample data, such as Fig.12 As shown, the preset threshold is set to 0.38. The methylation beta value of the site is greater than or equal to 0.38 and is classified as thalamic-like, while the methylation beta value is less than 0.38 and is classified as pons-like.

[0061] In one embodiment, S21 also includes position features, acquiring clinical data, extracting position features based on the clinical data to obtain position features, inputting methylation features and position features into a trained classifier for classification to obtain classification results of thalamic-like tumors or pontine-like tumors.

[0062] In one embodiment, S2 is replaced by S22: extracting gene features from the tumor omics data to obtain gene features; the method further includes S32: inputting the gene features into a trained classifier for classification to obtain classification results of thalamic-like tumors or pons-like tumors.

[0063] In one embodiment, the gene signature is the MEX3A gene expression value.

[0064] In one embodiment, S22 also includes position features, acquiring clinical data, extracting position features based on the clinical data to obtain position features, inputting gene features and position features into a trained classifier for classification to obtain classification results of thalamic-like tumors or pons-like tumors.

[0065] In one embodiment, the S22 also includes methylation features, and methylation features are extracted from the tumor omics data to obtain methylation features, and the gene features and methylation features are input into a trained classifier for classification to obtain classification results of thalamic-like tumors or pontine-like tumors.

[0066] In one embodiment, the S22 also includes methylation features and position features; extracting methylation features from the tumor omics data to obtain methylation features; acquiring clinical data, and extracting position features based on the clinical data to obtain position features; inputting the gene features, methylation features, and position features into a trained classifier for classification to obtain classification results of thalamic-like tumors or pontine-like tumors.

[0067] In one embodiment, S2 is replaced by S23: obtaining tumor cell immune infiltration data, calculating the positive number of CD14 and / or CD8 in the cell immune infiltration data, and when the positive number of CD14 and / or CD8 is greater than a preset threshold, it is determined to be a thalamic-like tumor; when the positive number of CD14 and / or CD8 is less than the preset threshold, it is determined to be a pons-like tumor.

[0068] In one embodiment, the S20 also includes immune infiltration features, obtaining tumor cell immune infiltration data, calculating the CD14 and / or CD8 positive counts in the cell immune infiltration data, and inputting the mutation features and immune infiltration features into a trained classifier for classification to obtain the classification results of thalamic-like tumors or pons-like tumors.

[0069] In one embodiment, the S20 also includes methylation features, gene features, and immune infiltration features. Gene features are extracted from the tumor omics data to obtain gene features, methylation features are extracted from the tumor omics data to obtain methylation features, tumor cell immune infiltration data is obtained, and the CD14 and / or CD8 positive counts in the cell immune infiltration data are calculated. The mutation features, methylation features, gene features, and immune infiltration features are input into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pons-like tumor.

[0070] In one embodiment, the S20 also includes methylation features, and the methylation features are extracted from the tumor omics data to obtain methylation features, and the mutation features and methylation features are input into a trained classifier for classification to obtain the classification results of thalamic-like tumors or pontine-like tumors.

[0071] In one embodiment, the S20 also includes gene features, and gene features are extracted from the tumor omics data to obtain gene features. The mutation features and gene features are input into a trained classifier for classification to obtain classification results of thalamic-like tumors or pontine-like tumors.

[0072] In one embodiment, the S20 also includes methylation features and gene features, and gene features are extracted from the tumor omics data to obtain gene features. The mutation features, methylation features, and gene features are input into a trained classifier for classification to obtain classification results of thalamic-like tumors or pons-like tumors.

[0073] In one embodiment, the S20 also includes methylation features, gene features, and position features; gene features are extracted from the tumor omics data to obtain gene features; clinical data are acquired, and position features are extracted based on the clinical data to obtain position features; the mutation features, methylation features, gene features, and position features are input into a trained classifier for classification to obtain classification results of thalamic-like tumors or pontine-like tumors.

[0074] In a specific embodiment, the present invention analyzes multi-omics and spatially resolved single-cell data from 36 H3-DSGs and finds two different subgroups: pontine-like and thalamic-like. Thalamic-like tumors are characterized by NF1 mutations and show more immune infiltration, including a large number of proinflammatory myeloid cells and CD8+ cytotoxic T cells. The age of patients in this subtype group is relatively high and the prognosis is relatively good. In contrast, pontine-like tumors show abundant TP53 mutations, high proliferation niches, and genomic instability. The age of patients in this subtype group is relatively low and the prognosis is relatively poor.

[0075] In a specific embodiment, the present invention found two subtypes of spinal glioma with H3 K27M mutation, with different prognosis, age distribution, and mutation spectrum, respectively simulating the mutation characteristics of gliomas in different anatomical locations in the brain. In order to clarify the subtypes of spinal glioma with H3 K27M mutation, the present invention performed multi-omics clustering for subtype identification using 36 cases with DNA methylation, WES and RNA sequencing data ( Figure 4 B). There were 22 males and 14 females, ranging in age from 7 to 61 years old, with a median age of 33 years old. The tumors were extensively resected from the first cervical segment (C1) to the first lumbar segment (L1). Two of the cases had multifocal tumors.

[0076] In a specific embodiment, in order to determine the key somatic mutations, the present invention identified key mutation features based on the SAVI2 and MutsigCV algorithms for 36 patients in the discovery cohort. Interestingly, the present invention found 2 cases in the discovery cohort that exhibited H3 K27I mutations, and all other cases exhibited H3K27M mutations. In addition to H3 K27 mutations, the present invention also detected previously reported mutations in brain H3 K27 mutant dmgs, including mutations in TP53 (16 / 36), PPM1D (13 / 36), NF1 (15 / 36), ATRX (5 / 36), ACVR1 (3 / 36) and PIK3R1 (2 / 36) ( Figure 4 B).

[0077] In one embodiment, the present invention develops a network-based multi-omics integration method named constrained similarity network fusion (cSNF, Figure 5 cSNF first constructs a patient similarity network from each omics data, fuses these multi-layer networks to represent the comprehensive associations between patients, and finally clusters the patients using a graph-based spectral clustering method. Using cSNF, the present invention stratifies 36 H3-DSGs into two subtypes with different clinical and genetic characteristics ( Figure 4 C). Patients with subtype 1 had significantly shorter overall survival (P = 4.1e-04) and were diagnosed at a younger age (P = 4.5e-05) compared with subtype 2. The median age of patients with type 1 was 18 years and the median overall survival was 11.9 months, while the median age of patients with type 2 was 37.5 years and the median overall survival was 27.6 months ( Figure 4 C and Figure 5 cSNF showed superior statistical power in terms of overall survival and age stratification compared with single-omics clustering or voting by single-omics clustering ( Figure 5Of note, we found that TP53 mutations were enriched in subtype 1 (13 / 16 vs. 3 / 20, P = 1.25e-4), while NF1 mutations were enriched in subtype 2 (1 / 16 vs. 13 / 20, P = 4.31e-4, Figure 4 B and Figure 4 C).

[0078] In a specific embodiment, the cSNF method: a constrained similarity network fusion typing method based on multi-omics data, the method includes three steps: constructing a multi-omics patient-patient similarity network (PPSN), fusing constrained similarity networks and spectrum-based clustering.

[0079] First, the present invention uses the weighted technology in the Similarity Network Fusion (SNF) algorithm to select these features from different omics data to construct PPSNs. The algorithm first calculates the similarity between two patients and then determines the weight of the edge between them by scaling the exponential similarity kernel. Then, according to the SNF algorithm definition, the normalized weight matrix (P) and local affinity matrix (A) of each PPSN are constructed. SNF further uses iterative fusion and normalized embedding technology to integrate multi-omics data. In this iterative fusion process, the present invention has been modified as shown below:

[0080] Among them, v represents one of the different omics data, such as s represents somatic cells, c represents CNV, m represents methylation, and r represents RNA-seq. lab() is a data type function from different omics data, such as lab(s) and lab(c) are 'Genomics', lab(m) is 'Epigenomics', and lab(r) is 'Transcriptomics'. N represents the unique number of data types, T represents transposition, and k represents the omics data label. k=1, Z(k) represents s; k=2, Z(k) represents c; k=3, Z(k) represents m; k=4, Z(k) represents r.

[0081] The modified approach is that information belonging to the same data type is considered parallel and not integrated at the same label level.

[0082] Finally, after a certain iterative process, the final fused PPSN can calculate the mean of the fused normalized weight matrix. The graph-based spectral clustering method is further used to find subgroups on the fused cluster nodes, and the number of clusters is estimated by optimizing RatioCut.

[0083] In a specific embodiment, to better characterize the tumor characteristics of H3-DSG subtype 1 and subtype 2, the present invention compared the somatic mutation spectrum, age at diagnosis, and overall survival outcomes of the H3-DSG subtypes with those of previously reported brain DMG subtypes in an independent cohort consisting of m-pontine and m-medullary brain DMG17. Tumors of the m-pontine DMG subtype are mainly located in the pons, while tumors of the m-medullary DMG subtype are mostly located outside the pons. Strikingly, the newly discovered H3-DSG Subtype 1 exhibited somatic characteristics similar to those of the m-pontine DMG subtype, while the somatic characteristics of H3-DSG Subtype 2 were similar to those of the m-medullary brain DMG subtype, with the only exceptions being SF3B1 and FGFR1 mutations, which were more frequent in m-medullary brain DMGs than in H3-DSG Subtype 2 ( Figure 4 D). The age distribution at diagnosis was similar in the H3-DSG Subtype 1 and M-Pons brain DMG groups and was significantly lower than that in the H3-DSG Subtype 2 (P=4.5e-05) and M-Pons brain DMG groups (P= 6.4e-06), respectively ( Figure 4 E). The survival rates of the H3-DSG 1 group and the m-pontine DMG group were significantly lower than those of the H3-DSG 2 group (P = 8.7e-04) and the m-medullary DMG group (P = 1.4e-04), respectively ( Figure 4 F). These results indicate that H3-DSG subtype 1 and H3-DSG subtype 2 have similar molecular and clinical characteristics to m-pontine and m-medullary DMG, respectively. Based on these findings, the present invention designates H3-DSG subtype 1 as pontine-like and H3-DSG subtype 2 as thalamic-like.

[0084] In a specific embodiment, transcriptome, single cell and protein multicolor staining clearly show that pontine-like H3-DSG has higher proliferation activity and DNA instability, while thalamic-like H3-DSG has higher immune infiltration.

[0085] Specifically, by investigating the differentially expressed genes between pontine-like and thalamic-like H3-DSGs ( Figure 6 A), the present invention found that the genes significantly upregulated in pontine-like tumors are mainly related to cell division and cell cycle-related functions, such as MEX3A, KIF2C, FANCI, etc. ( Figure 6 A and Figure 7 In contrast, genes upregulated in thalamic-like tumors included immune and inflammatory activities ( Figure 7Thus, thalamic-like tumors had lower tumor purity (P = 0.006) and higher lymphocyte (P = 0.0044) and myeloid scores (P = 0.0006) compared with pontine-like H3-DSGs ( Figure 6 BD). In fact, scRNA-seq data showed that H3-DSGs of pontine-like and thalamic-like tumors showed different cell type compositions and tumor cell states. Malignant cells accounted for 80% to 90% of pontine-like tumors, and the remaining components were mostly myeloid cells, with few T cells. In contrast, in thalamic-like tumors, malignant cells accounted for only about 50%, and the remaining 50% included myeloid cells, T cells, NK cells, neutrophils, endothelial cells, and oligodendrocytes ( Figure 7 C). Consistent with this, immunohistochemistry (IHC) of 28 H3-DSGs showed that CD14-positive myeloid cells were more abundant in thalamic-like H3-DSGs ( Fig.14 ).

[0086] We also showed through multicolor staining that myeloid and lymphoid cells were more abundant in thalamic-like H3-DSGs. Correspondingly, CD14 (myeloid), CD3e (T cells), MPO (neutrophils), CD44, HLA-A and HLA-DR (antigen presentation) were more abundant in thalamic-like H3-DSGs than in pontine-like H3-DSGs ( Figure 8 ).

[0087] In one embodiment, the single cell analysis results also showed that the proportion of CD8-positive cytotoxic T cells in T cells was significantly increased in the thalamic-like H3-DSG ( Fig. 9 C). The results of immunohistochemistry showed that the proportion of CD3e-positive cells and GZMB-positive cells in thalamus-like H3-DSG was also significantly higher than that of the corresponding cells in pons-like H3-DSG ( Fig. 9 D). This indicates an increase in the proportion of thalamic-like H3-DSG-like cytotoxic T cells.

[0088] In a specific embodiment, MEX3A methylation sites were used: based on a discovery data set containing 36 samples, a naive Bayesian algorithm was used to build the model, and the Leave-One-Out Cross-Validation (LOOC) strategy was used to evaluate the model performance. The model showed good classification ability, achieving a classification accuracy of 97.2%. At the same time, its AUC value was as high as 0.984, further confirming the efficiency and stability of the model. The ROC curve is shown in the figure below. Fig.10 shown.

[0089] In a specific embodiment, using MEX3A gene expression value: based on a discovery data set containing 36 samples, the naive Bayes algorithm was used to build the model, and the Leave-One-Out Cross-Validation (LOOCV) strategy was used to evaluate the model performance. The model showed good classification ability, reaching a classification accuracy of 88.9%. At the same time, its AUC value was as high as 0.968, further confirming the efficiency and stability of the model. The ROC curve is shown in the figure below. Fig.11 shown.

[0090] The disclosed embodiments of the present invention further provide a computer program product or system, including a computer program, which, when executed by a processor, implements the above-mentioned steps of the method for classifying gliomas based on H3 K27 mutations.

[0091] Figure 2 A schematic diagram of a classification system for gliomas based on H3 K27 mutations provided in an embodiment of the present invention specifically includes: Acquisition unit: acquiring mutation data of a sample to be tested, where the sample to be tested is a glial tumor sample including H3 K27 mutation; A judgment unit: classifies the mutation data to obtain a classification result of a thalamic-like tumor or a pontine-like tumor; wherein, when the mutation feature is an NF1 mutation, it is judged to be a thalamic-like tumor; when the mutation feature is a TP53 mutation, it is judged to be a pontine-like tumor.

[0092] Figure 3 A schematic diagram of a glioma classification device based on H3 K27 mutation provided in an embodiment of the present invention specifically includes: A memory and a processor; the memory is used to store program instructions; the processor is used to call the program instructions, and when the program instructions are executed, any one of the above-mentioned glioma classification methods based on H3 K27 mutations is implemented.

[0093] The disclosed embodiments of the present invention further provide a computer-readable storage medium storing a computer program, wherein the computer program is executed by a processor to perform any of the above-mentioned methods for classifying gliomas based on H3 K27 mutations.

[0094] The verification results of this verification embodiment show that assigning inherent weights to indications can improve the performance of the method relative to the default setting. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the coupling or direct coupling or communication connection between each other shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms. The unit described as a separate component may or may not be physically separated, and the component displayed as a unit may or may not be a physical unit, that is, it may be located in one place, or it may be distributed on multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units. A person of ordinary skill in the art may understand that all or part of the steps in the various methods of the above-mentioned embodiments may be completed by instructing the relevant hardware through a program, and the program may be stored in a computer-readable storage medium, and the storage medium may include: a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, etc.

[0095] A person skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be implemented by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. The above-mentioned medium storage can be a read-only memory, a disk or an optical disk, etc.

[0096] The above is a detailed introduction to a computer device provided by the present invention. For a person skilled in the art, according to the concept of the embodiments of the present invention, there may be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for classifying gliomas based on H3 K27 mutations, characterized in that: include: S1. Obtain mutation data of a sample to be tested, where the sample to be tested is a glial tumor sample including an H3 K27 mutation; S2. Classify the mutation data to obtain a classification result of a thalamic-like tumor or a pontine-like tumor; wherein, when the mutation feature is an NF1 mutation, it is determined to be a thalamic-like tumor, and when the mutation feature is a TP53 mutation, it is determined to be a pontine-like tumor.

2. The method for classifying gliomas based on H3 K27 mutation according to claim 1, characterized in that: The S2 also includes tumor location extraction, obtaining clinical data, extracting location features based on the clinical data to obtain location features, and determining that the tumor is a thalamic-like tumor when the mutation feature is an NF1 mutation and the tumor is located outside the pons; and determining that the tumor is a pons-like tumor when the mutation feature is a TP53 mutation and the tumor is located inside the pons; Optionally, the classification is to cluster the mutation data to obtain a clustering result, and when the clustering feature includes NF1 mutation, it is determined to be a thalamic-like tumor, and when the clustering feature includes TP53 mutation, it is determined to be a pontine-like tumor.

3. The method for classifying gliomas based on H3 K27 mutation according to claim 1, characterized in that: The classification is performed by a trained multi-omics integration model, and the training process of the multi-omics integration model is: Obtain patient omics datasets; constructing a patient similarity network set based on the omics dataset; The distances between the similarity networks of different patients are calculated to obtain the weights of the edges between the similarity networks; Constructing a normalized weight matrix and a local affinity matrix of each similarity network based on the weights of the edges between the similarity networks; Iteratively fusing the normalized weight matrix and the local affinity matrix to obtain a fused matrix; Clustering the fused matrix to obtain a clustering result; Optionally, the training process further includes a normalization calculation, performing a normalization calculation on the fused matrix to obtain a fused normalized weight matrix, and clustering the fused normalized weight matrix to obtain a clustering result; Optionally, the fused normalized matrix further includes mean calculation, performing an average calculation on the fused normalized matrix to obtain a mean of the fused normalized matrix, and clustering the mean of the fused normalized matrix to obtain a clustering result; Optionally, the omics data includes one or more of the following: somatic mutation data, copy number variation data, methylation data, transcriptome data; Optionally, the clustering includes: a graph-based spectral clustering algorithm; Optionally, the iterative fusion includes performing mean calculation on normalized weight matrices of different omics data types to obtain an average normalized matrix, and the average normalized matrix is ​​calculated with a local affinity matrix to obtain a fused matrix; Optionally, the iterative fusion process is expressed as: in, v Represents one of the different omics data, such as s Represents somatic cells, c Represents CNV, m Represents methylation, r represents RNA-seq, A represents the local affinity matrix, and P represents the normalized weight matrix. lab () is a function of the data type from different omics data, N A unique number representing a data type, T represents transpose, k Indicates the omics data label, k =1, Z( k )express s ; k =2, Z( k )express c ; k =3, Z( k )express m ; k =4, Z( k )express r .

4. The method for classifying gliomas based on H3 K27 mutation according to claim 1, characterized in that: The S2 is replaced by S20: acquiring tumor omics data, and extracting mutation features from the tumor omics data to obtain mutation features; the method further comprises S30: inputting the mutation features into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pontine-like tumor; Optionally, the classifier includes one or more of the following: convolutional neural network, Transformer, residual network, hole convolutional neural network, extreme learning machine, support vector machine, XGBoost, AdaBoost, random forest, naive Bayes.

5. The method for classifying gliomas based on H3 K27 mutation according to claim 1, characterized in that: The S2 is replaced by S21: acquiring tumor omics data, and extracting methylation features from the tumor omics data to obtain methylation features; the method further comprises S31: inputting the methylation features into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pontine-like tumor; Optionally, the methylation feature is a MEX3A methylation site; Optionally, the S31 is replaced by: performing classification based on the methylation feature to obtain a classification result of a thalamic-like tumor or a pontine-like tumor, wherein the methylation feature is a MEX3A methylation site, and when the MEX3A methylation site value is greater than or equal to a preset threshold, it is determined to be a thalamic-like tumor, and when the MEX3A methylation site value is less than the preset threshold, it is determined to be a pontine-like tumor; Optionally, S21 also includes position features, acquiring clinical data, extracting position features based on the clinical data to obtain position features, inputting methylation features and position features into a trained classifier for classification to obtain classification results of thalamic-like tumors or pontine-like tumors.

6. The method for classifying gliomas based on H3 K27 mutation according to claim 1, characterized in that: The S2 is replaced by S22: acquiring tumor omics data, and extracting gene features from the tumor omics data to obtain gene features; the method further comprises S32: inputting the gene features into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pontine-like tumor; Optionally, the gene feature is a MEX3A gene expression value; Optionally, the S22 further includes position features, obtaining clinical data, extracting position features based on the clinical data to obtain position features, inputting gene features and position features into a trained classifier for classification to obtain a classification result of thalamic-like tumors or pontine-like tumors; Optionally, the S22 further includes methylation features, extracting methylation features from the tumor omics data to obtain methylation features, inputting the gene features and methylation features into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pontine-like tumor; Optionally, the S22 further includes methylation characteristics and position characteristics; Extracting methylation features from the tumoromics data to obtain methylation features; acquiring clinical data, and extracting position features based on the clinical data to obtain position features; inputting the gene features, methylation features, and position features into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pontine-like tumor; Optionally, S2 is replaced by S23: obtaining tumor cell immune infiltration data, calculating the positive number of CD14 and / or CD8 in the cell immune infiltration data, and when the positive number of CD14 and / or CD8 is greater than a preset threshold, determining it as a thalamic-like tumor; when the positive number of CD14 and / or CD8 is less than the preset threshold, determining it as a pons-like tumor.

7. The method for classifying gliomas based on H3 K27 mutation according to claim 4, characterized in that: The S20 further includes methylation features, extracting methylation features from the tumor omics data to obtain methylation features, and inputting the mutation features and methylation features into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pontine-like tumor; Optionally, the S20 further includes gene features, extracting gene features from the tumor omics data to obtain gene features, and inputting the mutation features and gene features into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pontine-like tumor; Optionally, the S20 further includes immune infiltration features, obtaining tumor cell immune infiltration data, calculating the number of CD14 and / or CD8 positive cells in the cell immune infiltration data, and inputting the mutation features and immune infiltration features into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pontine-like tumor; Optionally, the S20 further includes methylation features and gene features, extracting gene features from the tumor omics data to obtain gene features, extracting methylation features from the tumor omics data to obtain methylation features, inputting the mutation features, methylation features, and gene features into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pontine-like tumor; Optionally, the S20 further includes methylation features, gene features, and immune infiltration features, extracting gene features from the tumor omics data to obtain gene features, extracting methylation features from the tumor omics data to obtain methylation features, acquiring tumor cell immune infiltration data, calculating the number of CD14 and / or CD8 positive cells in the cell immune infiltration data, and inputting the mutation features, methylation features, gene features, and immune infiltration features into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pontine-like tumor; Optionally, the S20 further includes methylation features, gene features, and position features; Extracting gene features from the tumoromics data to obtain gene features; acquiring clinical data, and extracting position features based on the clinical data to obtain position features; inputting the mutation features, methylation features, gene features, and position features into a trained classifier for classification to obtain a classification result of a thalamic-like tumor or a pontine-like tumor; Optionally, the S20 also includes methylation characteristics, gene characteristics, position characteristics, and immune infiltration characteristics; gene characteristics are extracted from the tumor omics data to obtain gene characteristics; clinical data are obtained, and position characteristics are extracted based on the clinical data to obtain position characteristics; tumor cell immune infiltration data are obtained, and the number of CD14 and / or CD8 positive cells in the cell immune infiltration data is calculated; the mutation characteristics, methylation characteristics, gene characteristics, position characteristics, and immune infiltration characteristics are input into a trained classifier for classification to obtain the classification results of thalamic-like tumors or pons-like tumors.

8. A computer program product comprising a computer program or instructions, characterized in that: The computer program or instructions are executed by a processor to implement the method for classifying gliomas based on H3 K27 mutations according to any one of claims 1 to 7.

9. A computer device comprising a memory, a processor and a computer program or instruction stored in the memory, characterized in that: The computer program or instructions are executed by a processor to implement the method for classifying gliomas based on H3 K27 mutations according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program or instruction stored thereon, characterized in that: The computer program or instructions are executed by a processor to implement the method for classifying gliomas based on H3 K27 mutations according to any one of claims 1 to 7.

Images