Breast cancer cell group identification and metastatic or drug-resistant subtype tumor cell feature identification method and system

Through correlation calculation and marker gene weight construction with breast cancer histocyte-specific marker dataset, the accuracy and efficiency of cell taxa identification in single-cell transcriptome sequencing data were solved, and the tumor cells with metastasis or drug-resistant properties of breast cancer were identified , provides an accurate breast cancer cell population identification and feature recognition system.

CN118942555AActive Publication Date: 2024-11-12HUAZHONG UNIV OF SCI & TECH

Patent Information

Application Number
CN202410966873.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-18
Publication Date
2024-11-12
Estimated Expiration
2044-07-18

AI Technical Summary

Technical Problem

In the existing single-cell transcriptome sequencing data, the cell taxa identification method has problems such as inaccurate classification, high classification cost and inconsistent classification standards, and it is difficult to accurately identify tumor cells with breast cancer metastasis or drug-resistant characteristics.

Method used

The correlation calculation is performed by calculating the clustered cell population and each cell type in the breast cancer histocyte-specific labeling dataset, and the cell type corresponding to each clustered cell population is determined, and a breast cancer histocyte-specific labeling dataset is constructed. The marker is used Gene weights were annotated for cell taxa to identify epithelial stromal transformation and taxane drug resistance characteristic tumor cells.

Benefits of technology

It solves the problems of inaccurate classification, high cost and inconsistent standards in cell taxa identification, and provides a cell taxa annotation based on single-cell transcriptome sequencing data of breast cancer and a subpopulation of EMT or taxa resistance. The visualization system of molecular feature recognition improves the accuracy and efficiency of recognition.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118942555A_ABST
    Figure CN118942555A_ABST
Patent Text Reader

Abstract

The invention relates to a breast cancer cell group identification method and system and a metastatic or drug-resistant subtype tumor cell feature identification method and system, and belongs to the technical field of biological information analysis. Carrying out dimensionality reduction and clustering on the single cell sequencing expression matrix data to obtain a clustered cell population; and performing correlation calculation on the obtained clustered cell population and each cell type in the breast cancer tissue cell specific labeling data set, and judging the cell type corresponding to each clustered cell population. According to the method, the problems of inaccurate classification, relatively high classification cost and non-uniform classification standards in the prior art are solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of bioinformatics analysis, and more specifically, to a method and system for identifying breast cancer cell populations and metastatic or drug-resistant subtype tumor cell characteristics. Background Art

[0002] With the development of biotechnology, single-cell transcriptome sequencing technology can now perform high-throughput sequencing of transcriptomes at the single-cell level, and a single sequencing can obtain transcriptome expression data of thousands to tens of thousands of cells. Due to the heterogeneity and complexity between cells in breast cancer tissue, single-cell transcriptome sequencing is widely used in breast cancer. By analyzing the single-cell transcriptome sequencing data of breast cancer tissue, the molecular characteristics and biological functions of different cells in the tumor tissue of breast cancer patients can be revealed, which is conducive to exploring the potential pathogenesis and personalized treatment targets of patients.

[0003] According to data from the International Agency for Research on Cancer and the World Health Organization in 2023, there will be 2,752,000 new cases of breast cancer and 684,996 deaths from breast cancer worldwide in 2023. Tumor metastasis and treatment resistance are the main causes of disease progression and death in breast cancer patients. Epithelial-mesenchymal transition (EMT) is a biological process in which epithelial cells lose polarity and intercellular connections, allowing breast cancer cells to have the ability to metastasize from the primary lesion to distant organs, and is an important process in tumor metastasis. Taxanes inhibit the division and proliferation of cancer cells by blocking the polymerization of microtubules in cells. They are widely standardized and included in the standard treatment of breast cancer, especially for the treatment of locally advanced or metastatic breast cancer. However, patients often develop resistance to taxanes during treatment, which affects the treatment effect and prognosis.

[0004] Through single-cell transcriptome sequencing, the molecular characteristics of tumor subpopulations with metastasis or drug resistance in breast cancer patients can be identified, and the differentially expressed genes of metastatic or drug-resistant tumor cells and their enriched biological processes can be obtained, which can guide treatment strategies, more accurately select appropriate treatment options, and reduce the risks of metastasis and drug resistance. The basic process of single-cell transcriptome sequencing data analysis includes sequencing data quality control, data standardization, identification of highly variable genes, data dimensionality reduction, cell clustering, cell group identification, determination of differential genes in characteristic cell subpopulations, and analysis of differential gene enrichment pathways. Among them, cell group identification is a key step. Incorrect cell group identification will lead to incorrect judgments on the function, state, and physiological significance of such cells, misleading the subsequent biological research direction and conclusions.

[0005] At present, the most commonly used method for identifying cell clusters in single-cell transcriptome sequencing data is a standardized classification method based on reference databases and machine learning. However, there are some problems with this method. First, it depends on the coverage and quality of the reference database. The reference database cannot cover all cell types or newly discovered cell subtypes, and the annotation accuracy of rare cell types is low. Secondly, the choice of machine learning model structure and parameters has a significant impact on the results. A large amount of labeled data is required to train the model, and the threshold for use is high. Another commonly used method for identifying cell clusters in single-cell transcriptome sequencing data is to manually associate each cell cluster with a known cell type based on cell-specific marker genes. This method also has problems. First, cell-specific marker genes often rely on other studies and have certain limitations; secondly, there are often noise signals and rare cell types in single-cell sequencing data, which are easily ignored or confused with other types in cluster analysis; and, manually annotated cell types are easily affected by the subjective understanding and experience of researchers, and it is difficult to define the expression level of cell-specific marker genes. The steps are cumbersome, time-consuming, and inefficient.

[0006] In summary, the existing cell population identification methods have problems such as inaccurate classification, high classification cost, and inconsistent classification standards. The exploration of the pathogenesis of breast cancer patients and the personalized diagnosis and treatment process require the identification of molecular characteristics of tumor subpopulations with metastasis or drug resistance. Summary of the invention

[0007] In order to overcome the deficiencies of existing cell group identification technology and the need for molecular feature identification of tumor cells based on breast cancer metastasis or drug resistance characteristics, the purpose of the present invention is to provide a cell group identification of breast cancer single cell transcriptome sequencing data, as well as epithelial mesenchymal transition (EMT) or taxane drug resistance characteristic subtype tumor cell identification method and system. The present invention performs correlation calculations on each cell type in the clustered cell group and the breast cancer tissue cell-specific marker data set to determine the cell type corresponding to each clustered cell group. The present invention solves the problems of inaccurate classification, high classification cost, and inconsistent classification standards in the existing cell group identification technology, and provides a visualization system for annotation of cell groups based on breast cancer single cell transcriptome sequencing data and molecular feature identification of tumor subgroups characterized by EMT or taxane drug resistance.

[0008] According to a first aspect of the present invention, a method for identifying a breast cancer cell population is provided, comprising the following steps:

[0009] (1) Perform dimensionality reduction and clustering on single-cell sequencing expression matrix data to obtain clustered cell populations;

[0010] (2) Calculate the correlation between the clustered cell groups obtained in step (1) and each cell type in the breast cancer tissue cell-specific marker data set to determine the cell type corresponding to each clustered cell group.

[0011] Preferably, the breast cancer tissue cell-specific marker dataset is constructed by the following steps:

[0012] S1: Obtaining marker genes of basic cells of breast cancer tissue, wherein the basic cells of breast cancer tissue are 15 types, namely, epithelial cells, endothelial cells, fibroblasts, tumor-associated fibroblasts, T cells, Treg cells, B cells, NK cells, NKT cells, macrophages, monocytes, dendritic cells, mast cells, tumor cells and tumor stem cells;

[0013] S2: sorting the marker genes of each basic cell obtained in step S1, and sorting them from high to low importance according to the ranking of the marker genes in the CellMarker2.0 database, and obtaining the top 10 marker genes of each basic cell;

[0014] S3: The weight of each marker gene of each basic cell is obtained by calculating the number of times each marker gene of each basic cell appears in the CellMarker2.0 database / the total number of times the top 10 marker genes of the basic cell appear in the CellMarker2.0 database;

[0015] S4: The basic cells of the breast cancer tissue and the marker genes of each basic cell obtained in step S1, as well as the weights of each marker gene of each basic cell obtained in step S3, are used to construct a breast cancer tissue cell-specific marker dataset.

[0016] Preferably, the step (2) specifically comprises the following steps:

[0017] S1: finding the top 10 marker genes of the 15 basic cells in the breast cancer tissue cell-specific marker dataset in the clustered cell population, and obtaining the average expression level of each marker gene in each clustered cell population;

[0018] S2: Normalize the average expression levels of the top 10 marker genes of the 15 basic cell types in each clustered cell group to obtain the expression scores of the top 10 marker genes of each basic cell type in each clustered cell group;

[0019] S3: For each cell type, the expression scores of the top 10 marker genes in each clustered cell group are multiplied by their weights, and then summed to obtain the classification score of each cell group;

[0020] S4: The cell type corresponding to the highest classification score of each clustered cell group is marked as the cell type of the respective clustered cell group.

[0021] Preferably, quality control and standardization processing are also included before the dimensionality reduction.

[0022] According to another aspect of the present invention, a method for identifying characteristics of breast cancer cell metastasis or drug-resistant subtype tumor cells is provided, comprising the following steps:

[0023] (1) Calculate the relative expression of epithelial-mesenchymal transition marker genes or taxane drug marker genes in breast cancer tumor cell populations;

[0024] (2) extracting cells with a relative expression of epithelial-mesenchymal transition marker genes greater than 0 and setting them as epithelial-mesenchymal transition marker gene characteristic tumor cells, extracting cells with a relative expression of epithelial-mesenchymal transition marker genes less than 0 and setting them as non-epithelial-mesenchymal transition marker gene characteristic tumor cells; extracting cells with a taxane resistance expression greater than 0 and setting them as taxane resistance characteristic tumor cells, and extracting cells with a taxane resistance expression less than 0 and setting them as non-taxane resistance characteristic tumor cells;

[0025] (3) Separating tumor cells with epithelial-mesenchymal transition and non-epithelial-mesenchymal transition characteristics, and obtaining violin plots of the epithelial-mesenchymal transition scores of the two; separating tumor cells with taxane resistance and non-taxane resistance characteristics, and obtaining violin plots of the taxane resistance scores of the two;

[0026] (4) Identify differentially expressed genes in tumor cells with epithelial-mesenchymal transition and non-epithelial-mesenchymal transition characteristics, and identify differentially expressed genes in tumor cells with taxane-resistant characteristics and taxane-non-resistant characteristics;

[0027] (5) Perform KEGG or GO enrichment analysis on several genes with significant differences among the differentially expressed genes obtained in step (4) to obtain enrichment pathways and visualized bubble diagrams.

[0028] According to another aspect of the present invention, there is provided a system for identifying breast cancer cell populations, comprising:

[0029] Obtaining a clustered cell population module: The clustered cell population module is used to reduce the dimension and cluster the single-cell sequencing expression matrix data;

[0030] Cell cluster identification module: The cell cluster identification module is used to calculate the correlation between the clustered cell clusters and each cell type in the breast cancer tissue cell-specific marker data set to determine the cell type corresponding to each clustered cell cluster.

[0031] According to another aspect of the present invention, a system for identifying characteristics of breast cancer cell metastasis or drug-resistant subtype tumor cells is provided, comprising:

[0032] Calculation module: The calculation module is used to calculate the relative expression of epithelial-mesenchymal transition marker genes or taxane drug marker genes in breast cancer tumor cell populations;

[0033] Extraction and setting module: The extraction and setting module is used to extract cells with a relative expression of epithelial-mesenchymal transition marker genes greater than 0 and set them as epithelial-mesenchymal transition marker gene characteristic tumor cells, extract cells with a relative expression of epithelial-mesenchymal transition marker genes less than 0 and set them as non-epithelial-mesenchymal transition marker gene characteristic tumor cells; extract cells with a taxane resistance expression greater than 0 and set them as taxane resistance characteristic tumor cells, and extract cells with a taxane resistance expression less than 0 and set them as non-taxane resistance characteristic tumor cells;

[0034] Obtaining score violin plot module: The obtaining score violin plot module is used to separate tumor cells with epithelial-mesenchymal transition characteristics and non-epithelial-mesenchymal transition characteristics, obtain violin plots of epithelial-mesenchymal transition scores for both, separate taxane-resistant and non-taxane-resistant tumor cells, and obtain violin plots of taxane-resistant scores for both;

[0035] Identification module: The identification module is used to identify differentially expressed genes in tumor cells with epithelial-mesenchymal transition and non-epithelial-mesenchymal transition characteristics, and to identify differentially expressed genes in tumor cells with taxane-resistant characteristics and taxol-non-resistant characteristics;

[0036] Enrichment module: The enrichment module is used to perform KEGG or GO enrichment analysis on several genes with significant differences among the differentially expressed genes obtained, and obtain enrichment pathways and visualized bubble diagrams.

[0037] In general, the above technical solution conceived by the present invention has the following technical advantages compared with the prior art:

[0038] The present invention provides a visualization system for annotation of cell groups and identification of molecular features of tumor subgroups with EMT or taxane drug resistance characteristics based on single-cell transcriptome sequencing data of breast cancer, which solves the problems of inaccurate classification, high classification cost and inconsistent classification standards in existing cell group identification technologies, and meets the needs of molecular feature identification of tumor cells with breast cancer metastasis or drug resistance characteristics. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A flow chart of a method for identifying cell populations based on single-cell transcriptome sequencing data of breast cancer and molecular signature recognition of tumor subpopulations with EMT or taxane drug resistance characteristics provided in an embodiment of the present invention.

[0040] Figure 2 This is a diagram of the internal structure of the system provided in an embodiment of the present invention.

[0041] Figure 3 This is the visualized web page display interface of the present invention. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0043] The purpose of the present invention is to provide a method and system for identifying cell populations based on single-cell transcriptome sequencing data of breast cancer and identifying subtype tumor cells with EMT or taxane drug resistance characteristics. The present invention solves the problems of inaccurate classification, high classification cost and inconsistent classification standards in existing cell population identification technologies, and provides a visualization system for cell population annotation based on single-cell transcriptome sequencing data of breast cancer and molecular feature identification of tumor subtypes with EMT or taxane drug resistance characteristics.

[0044] The present invention provides a method for identifying cell populations based on single-cell transcriptome sequencing data of breast cancer, comprising the following steps:

[0045] Step 1, obtaining breast cancer single-cell sequencing data, performing preprocessing, and obtaining a gene expression matrix;

[0046] Step 2, performing quality control, standardization, dimensionality reduction and clustering on the single-cell sequencing data to obtain clustered cell population data;

[0047] Step 3, performing correlation calculation between the clustered cell group data and each cell type in the constructed breast cancer tissue cell-specific marker data set to obtain the cell type closest to the clustered cell group;

[0048] Step 4, annotating cell groups based on the closest cell type;

[0049] Furthermore, in step 1, preprocessing includes quality control of raw sample data, alignment with the reference genome, quantitative analysis, and construction of a single-cell gene expression matrix;

[0050] Furthermore, in step 2, quality control includes setting a gene threshold of single cells to remove non-single cells and setting a mitochondrial percentage threshold of single cells to remove cells with poor activity;

[0051] Furthermore, the normalization process in step 2 includes eliminating noise caused by uneven sequencing depth, cell size differences, batch effects, etc., ensuring that the expression values ​​of different genes have similar numerical ranges and using the total gene expression of all cells to normalize the gene expression value of each single cell;

[0052] Furthermore, the dimensionality reduction and clustering method in step 2 includes reducing the dimension of the data by principal component analysis and finding genes whose expression changes significantly in different cell states, constructing cell neighbor relationships by calculating the Euclidean distance between each cell and all other cells, and dividing the cells into different groups according to the cell neighbor relationships and graph-based clustering algorithms, K-means clustering and spectral clustering methods;

[0053] Furthermore, in step 3, the breast cancer tissue cell-specific marker dataset includes:

[0054] The fifteen basic cell types of breast cancer tissue: epithelial cells, endothelial cells, fibroblasts, tumor-associated fibroblasts, T cells, Treg cells, B cells, NK cells, NKT cells, macrophages, monocytes, dendritic cells, mast cells, tumor cells and tumor stem cells;

[0055] Among the cell type-specific marker genes, the marker genes that appear in cancer tissue samples (except epithelial cells) are ranked from high to low according to the number of times they appear as cell type marker genes in the literature;

[0056] The top 10 marker genes for each cell type (if there are less than 10, replace them with NA);

[0057] The classification weight of each marker gene as a cell type (the weight is assigned according to the number of times the marker gene appears as a cell type marker gene in the literature);

[0058] Furthermore, in step 3, the process of calculating the correlation between the clustered cell population data and each cell type in the data set includes:

[0059] Search for the 1st to 10th marker genes of the fifteen basic cell types of breast cancer tissue in the data set in the clustered cell data frame;

[0060] Obtain the average expression levels of the 1st to 10th marker genes of the fifteen cell types in each cell population;

[0061] The average expression levels of the 1st to 10th marker genes of the fifteen cell types in each cell population were normalized to 0-1;

[0062] For each cell type, the expression scores of the 1st to 10th marker genes in each cell population were multiplied by their classification weights and summed;

[0063] Returns the classification scores of fifteen cell types for each cell population;

[0064] Furthermore, in step 4, the cell group annotation based on the closest cell type is to mark the cell type corresponding to the highest cell group classification score of each cell group as the cell type of the respective cell group.

[0065] The present invention provides a method for identifying molecular characteristics of tumor subpopulations with characteristics of epithelial-mesenchymal transition (EMT) of breast cancer tumor cells or taxane drug resistance, comprising the following steps:

[0066] Step 1, obtaining a tumor cell gene expression data frame from breast cancer single-cell sequencing data that has been identified as a cell population;

[0067] Step 2, grouping and clustering the above tumor cells;

[0068] Step 3, scoring the above tumor cells using tumor cell EMT marker genes and taxane drug resistance marker genes respectively;

[0069] Step 4, separating tumor cells with EMT characteristics and non-EMT characteristics, and separating taxane-resistant and non-taxane-resistant tumor cells;

[0070] Step 5, identifying the differentially expressed genes (DEGs) of EMT and non-EMT characteristic tumor cells and the DEGs of taxane-resistant and non-taxane-resistant tumor cells, respectively;

[0071] Step 6, enrichment analysis of the above DEGs was performed using the biological function annotation databases KEGG and GO respectively;

[0072] Furthermore, in step 1, the obtaining of the tumor cell gene expression data frame is to extract the tumor cells or tumor stem cell populations in the breast cancer single cell sequencing data frame that have been classified into tumor cells;

[0073] Furthermore, in step 2, the method of grouping and clustering tumor cells includes constructing a cell neighbor relationship by calculating the Euclidean distance between each tumor cell and all other tumor cells, and dividing the tumor cells into different groups according to the cell neighbor relationship and a graph-based clustering algorithm, K-means clustering and spectral clustering method;

[0074] Furthermore, in step 3, the tumor cell EMT or taxane drug resistance marker genes are collected and obtained through literature and the Cancer Therapy Response Gene Marker Database (CTR-DB);

[0075] Furthermore, the process of scoring the tumor cell population in step 3 includes:

[0076] Define the tumor cell EMT marker gene module and the tumor cell taxane resistance gene module;

[0077] The expression values ​​of 200 genes outside the module in tumor cells were randomly selected as background values, and the average expression values ​​of all genes in the module in tumor cells were calculated;

[0078] Further, in step 4, the EMT gene module scores higher than 0 are identified as EMT characteristic tumor cells, the EMT gene module scores lower than 0 are identified as non-EMT characteristic tumor cells, the taxane drug resistance gene module scores higher than 0 are identified as taxane drug-resistant tumor cells, and the taxane drug resistance gene module scores lower than 0 are identified as non-taxane drug-resistant tumor cells;

[0079] Preferably, the 200 DEGs with the most significant differences obtained in step 5 are screened for KEGG or GO analysis.

[0080] The present invention provides a breast cancer single-cell transcriptome sequencing data cell group identification and EMT or taxane drug-resistant subtype tumor cell feature recognition system, comprising:

[0081] Algorithm server and upload, parameter selection and display web page;

[0082] The algorithm server is used to store algorithms and calculate breast cancer single-cell transcriptome sequencing data cell populations and EMT or taxane drug resistance characteristic subtype tumor cell identification, to obtain breast cancer single-cell transcriptome sequencing cell population identification results and EMT or taxane drug resistance characteristic subtype tumor cell identification results;

[0083] The upload, parameter selection and display webpages are integrated on the same website, and are used to upload single-cell transcriptome sequencing data, select the result observation interface and display the calculation results respectively.

[0084] Furthermore, the algorithm server includes a data transmission and R script calling module, an R script calculation module and a result storage and calling module.

[0085] The data transmission and R script calling module is used to transmit the breast cancer single-cell sequencing data uploaded by the user to the backend, and call the calculation method of cell group identification and EMT or taxane-resistant subtype tumor cell feature recognition in the R script;

[0086] The R script computing module is used to identify cell populations and calculate the characteristics of EMT or taxane-resistant subtype tumor cells;

[0087] The result storage and call module is used to store R script calculation results and facilitate users to call result display files.

[0088] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0089] As attached Figure 1 As shown, the present invention provides a method for identifying cell groups based on single-cell transcriptome sequencing data of breast cancer and identifying tumor cells with EMT or taxane drug resistance characteristics subtype, comprising:

[0090] Step 1: Upload breast cancer single-cell sequencing expression matrix data to scBCcelltype software

[0091] In this embodiment, the supplementary file of single-cell sequencing data of breast tissue of breast cancer patients (GEO access number, GSM5354528) is first downloaded from the Gene Expression Omnibus (GEO) database, and decompressed to obtain three files: count_matrix_barcodes.tsv, count_matrix_genes.tsv, and count_matrix_sparse.mtx. Preferably, the three files are compressed and renamed to barcodes.tsv.gz, features.tsv.gz, and matrix.mtx.gz using 7-Zip software, and the three files are uploaded to the visualization web page.

[0092] Step 2: Quality control, standardization and dimensionality reduction clustering of the uploaded breast cancer single-cell sequencing expression matrix

[0093] In this embodiment, the main R package used is Seurat. First, CreateSeuratObject is used to create an analysis data frame named "Object". Non-single cells and cells with poor status are removed by setting the gene number threshold and mitochondrial gene percentage threshold in each cell. Preferably, the specific parameters are set to: nFeature_RNA>200&nFeature_RNA<6000&percent.mt<15&nCount_RNA>400. The single-cell gene expression level is standardized using the NormalizedData and ScaleData algorithms. The PCA algorithm is used to reduce the dimension of the data, the FindNeighbors algorithm is used to calculate the Euclidean distance between each cell and all other cells to construct the cell neighbor relationship, and the FindClusters algorithm is used to divide the cells into 8 cell groups, cluster0-7, and the UMAP (Uniform Manifold Approximation and Projection) or TSNE (t-Distributed Stochastic Neighbor Embedding) algorithm is used to obtain the distribution of cell groups in the UMAP or TSNE reduced dimension space.

[0094] Step 3: Determine the cell type of the clustered cell population

[0095] The construction process of the breast cancer tissue cell-specific marker dataset of the present invention is as follows:

[0096] The CellMarker2.0 database was used to search for marker genes of epithelial cells, endothelial cells, fibroblasts, tumor-associated fibroblasts, T cells, Treg cells, B cells, NK cells, NKT cells, macrophages, monocytes, dendritic cells, mast cells, tumor cells, and tumor stem cells in breast tissue;

[0097] Screening of marker genes for each cell type that appear in cancer tissue samples (except epithelial cells);

[0098] The marker genes of each cell type were ranked according to their ranking in the CellMarker2.0 database, and the ranking was obtained from high to low importance.

[0099] Select the top 10 marker genes in each cell type. If there are less than 10 marker genes in a cell type, replace them with NA.

[0100] The marker genes were weighted according to their occurrence times as cell type marker genes in the CellMarker2.0 database.

[0101] The construction process of the method for calculating the correlation between the clustered cell group data and each cell type in the above-constructed data set is as follows:

[0102] First, construct a cell group scoring function for clustered cell groups:

[0103] Set the input parameters of the function, including: the analysis data frame of cell clustering, the number of cell clusters, the cell type, the top1 marker gene (gene1), top2 marker gene (gene2) ... top10 (gene10) marker genes of the cell type, the weight of the top1 marker gene, the weight of the top2 marker gene ..., the weight of the top10 marker gene, specifically: data, cluster_num, celltype, gene1, gene2, gene3, gene4, gene5, gene6, gene7, gene8, gene9, gene10, weight1, weight2, weight3, weight4, weight5, weight6, weight7, weight8, weight9, weight10;

[0104] The gene1 score calculation for each cell population includes: searching for gene1 in the single-cell sequencing data. If gene1 is expressed, the average expression level of gene1 of the cell type in each cell population is calculated using a loop statement, and then the expression score of gene1 in each cell population is obtained by 0-1 normalization. If there is no expression of gene1, the expression score of gene1 is 0. Specifically, desired_feature<-gene1

[0105] if(desired_feature%in%rownames(data@assays$RNA@features)){

[0106] gene1<-VlnPlot(data,features=gene1,group.by="seurat_clusters",log=T,ncol=1)[[1]][["data"]]

[0107] gene1_score<-rep(0,cluster_num)

[0108] cell_count<-rep(0,cluster_num)

[0109] for(iin 1:nrow(gene1)){

[0110] gene1_score[as.numeric(gene1[i,][2])]=gene1_score[as.numeric(gene1[i,][2])]+as.numeric(gene1[i,][1])

[0111] cell_count[as.numeric(gene1[i,][2])]=cell_count[as.numeric(gene1[i,][2])]+1

[0112] }

[0113] gene1_score=gene1_score / cell_count

[0114] gene1_score_normalized=(gene1_score-min(gene1_score)) / (max(gene1_score)-min(gene1_score))

[0115] }else{

[0116] gene1_score_normalized=0

[0117] };

[0118] The process of calculating the scores of gene2 to gene10 in each cell population was the same as that of gene1;

[0119] The expression scores of gene1 to gene10 in each cell population are multiplied by their classification weights and summed to obtain the scores of the cell type in each cell population, specifically:

[0120] score=gene1_score_normalized*weight1+gene2_score_normalized*weight2+ge ne3_score_normalized*weight3+gene4_score_normalized*weight4+gene5_scor e_normalized*weight5+gene6_score_normalized*weight6+

[0121] gene7_score_normalized*weight7+gene8_score_normalized*weight8+

[0122] gene9_score_normalized*weight9+gene10_score_normalized*weight10;

[0123] Returns the score of this cell type for each cell population;

[0124] Next, use the above function to score the cell population, specifically:

[0125] celltype_score<-data.frame(marker_gene(Object,num_clusters,"Epithelial_cells","CDH1","CD117","CD166","K RT14","KRT18","DSP","EPCAM","GCDFP15","MCAM","TP63",7 / 16,1 / 16,1 / 16,1 / 16,1 / 16,1 / 16,1 / 16,1 / 16,1 / 16,1 / 16),

[0126] marker_gene(Object,num_clusters,"Endothelial_cells","PECAM1","CD31","VWF","IL3RA","CD34","CDH5","CLEC14A","EMCN","MCAM","PECAM 1",5 / 21,5 / 21,1 / 7,2 / 21,1 / 21,1 / 21,1 / 21,1 / 21,1 / 21,1 / 21),

[0127] marker_gene(Object,num_clusters,"Fibroblasts","DCN","PDGFRB","ACTA2","COL1A1","COL1A2"," ITGB1","PDGFRA","S100A4","TAGLN","THY1",1 / 6,1 / 6,1 / 12,1 / 12,1 / 12,1 / 12,1 / 12,1 / 12,1 / 12,1 / 12),

[0128] marker_gene(Object,num_clusters,"CAF","ACTA2","COL1A1","PDGFR A","PDGFRB","PDPN","FAP","THY1","NA","NA","NA",1 / 6,1 / 6,1 / 6,1 / 6,1 / 6,1 / 12,1 / 12,0,0,0),

[0129] marker_gene(Object,num_clusters,"T_cells","CD3D","CD3G","CD4","CD2","CD3","CD3E","CD8","CD137","CD28","CD8A",5 / 22,3 / 22,3 / 22,1 / 11,1 / 11,1 / 11,1 / 11,1 / 22,1 / 22,1 / 22),

[0130] marker_gene(Object,num_clusters,"Treg_cells","FOXP3","CCR8","IL2RA","BATF","CCR7","CD3D","CD3G","KLRB1","LAYN","MAGEH1",6 / 17,2 / 17,2 / 17,1 / 17,1 / 17,1 / 17,1 / 17,1 / 17,1 / 17,1 / 17),

[0131] marker_gene(Object,num_clusters,"B_cell","CD20","CD79A","MS4A1","CD19","BANK1","CD14","CD27","CD38","CD40","CD74",7 / 24,1 / 6,1 / 6,1 / 8,1 / 24,1 / 24,1 / 24,1 / 24,1 / 24,1 / 24),

[0132] marker_gene(Object,num_clusters,"NK_cell","KLRB1","KLRC1","KLRD 1","NKG7","CCR7","CD3D","CD3G","CD8A","GNLY","KIR2DL1",1 / 7,1 / 7,1 / 7,1 / 7,1 / 14,1 / 14,1 / 14,1 / 14,1 / 14,1 / 14),

[0133] marker_gene(Object,num_clusters,"NKT_cells","CD3D","GNLY","KLRD 1","NCR1","NA","NA","NA","NA","NA","NA",1 / 4,1 / 4,1 / 4,1 / 4,0,0,0,0,0,0),

[0134] marker_gene(Object,num_clusters,"Machrophage","CD68","CD14","CD163","CD209","CD33","CD4","CD86","CDH5","CSF1","FCGR3A",4 / 13,1 / 13,1 / 13,1 / 13,1 / 13,1 / 13,1 / 13,1 / 13,1 / 13,1 / 13),

[0135] marker_gene(Object,num_clusters,"Monocyte","CD14","CD33","CD4","FCGR3A","ITGAM","ITGAX","LAPTM5","LYZ","PTPRC","VIM",2 / 11,1 / 11,1 / 11,1 / 11,1 / 11,1 / 11,1 / 11,1 / 11,1 / 11,1 / 11),

[0136] marker_gene(Object,num_clusters,"DC","CD1C","HLA-DPA1","HLA-DPB1","HLA-DQB1","HLA-DRA","ITGAX","NRP1","NA","NA","NA",1 / 7,1 / 7,1 / 7,1 / 7,1 / 7,1 / 7,1 / 7,0,0,0),

[0137] marker_gene(Object,num_clusters,"Mast_cells","CAM1","CPA3","CTSG","KIT","TPSD1","TPSAB1","NA","NA","NA","NA",1 / 6,1 / 6,1 / 6,1 / 6,1 / 6,0,0,0,0,0),

[0138] marker_gene(Object,num_clusters,"cancer_stem_cells","CD44","CD24","ALDH1","CD133","ALDH","SOX2","EPCAM","POU5F1","BMI1","ITGA6",52 / 151,-29 / 151,22 / 151,13 / 151,8 / 151,8 / 151,6 / 151,6 / 151,4 / 151,3 / 151),

[0139] marker_gene(Object,num_clusters,"cancer_cells","ERBB2","EPCAM","GATA3","KRT18","ADH1B","AKR1A1","ALPI","ARHGAP10","B3GNT2","BC AM",5 / 17,2 / 17,2 / 17,2 / 17,1 / 17,1 / 17,1 / 17,1 / 17,1 / 17,1 / 17))

[0140] Next, the cell type corresponding to the highest cell group classification score of each cell population is marked as the cell type of the respective cell population, as follows:

[0141] Set the column name of the table obtained by scoring the cell groups in the previous step to the cell type name;

[0142] The cell type corresponding to the highest cell cluster score of each cell cluster is set as the cell type of the cell cluster, specifically: celltype = data.frame(cluster = c(0:(num_clusters-1)), celltype = apply(celltype_score, 1, function(t) colnames(celltype_score)[which.max(t)])), to obtain a cell cluster-cell type correspondence table;

[0143] In this embodiment, the cell types corresponding to the highest cell group classification scores of cell groups 0 to 7 are endothelial cells, tumor-associated fibroblasts, fibroblasts, monocytes, B cells, tumor cells, fibroblasts and NK cells, respectively.

[0144] Step 4: Obtain the distribution of cell populations in UMAP or TSNE reduced dimensionality space

[0145] In this embodiment, the cell group annotation information is added to the single-cell sequencing data frame, and the distribution of seven cell groups including tumor cells, endothelial cells, fibroblasts, etc. in the UMAP or TSNE dimensionality reduction space is obtained and output.

[0146] Step 5: Determine whether there are tumor cells in the cell cluster annotation results

[0147] In this embodiment, tumor cells exist, so the tumor cells are extracted and Step 6 to Step 10 are performed in sequence.

[0148] Step 6: Further group and cluster the tumor cells extracted above.

[0149] In this example, tumor cells are further grouped and clustered to obtain two tumor cell populations and their distribution in the UMAP or TSNE dimensionality reduction space.

[0150] Step 7: Use tumor cell EMT marker gene set or taxane drug resistance marker gene set to score tumor cells

[0151] Preferably, the tumor cell EMT marker gene set and the taxane drug resistance marker gene set are obtained from literature and the Cancer Therapy Response Gene Marker Database (CTR-DB) respectively.

[0152] In this embodiment, AddModuleScore in the Seurat package and the above-collected EMT and taxane resistance gene sets were used to calculate the EMT and taxane resistance scores of tumor cells, and EMT characteristic tumor cells, non-EMT characteristic tumor cells, taxane-resistant tumor cells, and non-taxane-resistant tumor cells were obtained respectively.

[0153] Step 8: Isolation of EMT-characteristic tumor cells and non-EMT-characteristic tumor cells, taxane-resistant tumor cells and non-taxane-resistant tumor cells

[0154] Preferably, cells with an EMT score greater than 0 are extracted and set as EMT characteristic tumor cells, cells with an EMT score less than 0 are extracted and set as non-EMT characteristic tumor cells, cells with a taxane resistance score greater than 0 are extracted and set as taxane resistance characteristic tumor cells, cells with a taxane resistance score less than 0 are extracted and set as non-taxane resistance characteristic tumor cells, and violin plots of tumor cell EMT scores and violin plots of taxane resistance scores are drawn respectively.

[0155] Step 9: Obtain differentially expressed genes in EMT and non-EMT characteristic tumor cells and differentially expressed genes in taxane-resistant and non-taxane-resistant tumor cells

[0156] In this embodiment, the FindMarkers function in the Seurat package is used. Preferably, the statistical test method used for comparison is "wilcox", genes expressed in at least 10% of the cells are retained, and genes with adjusted p-values ​​less than 0.05 are used as significantly differentially expressed genes, and a differentially expressed gene table is output.

[0157] Step 10: Pathway enrichment analysis of differentially expressed genes

[0158] In this embodiment, preferably, the 200 most significantly different genes are screened and KEGG and GO enrichment analysis are performed using enrichKEGG and enrichGO of clusterProfiler, respectively. The p-value threshold for screening significantly enriched pathways is set to 0.06, the q-value threshold for multiple testing correction is 0.05, and the method for multiple testing correction is Benjamini-Hochberg. The enriched pathway table and the visualized bubble chart are obtained and output.

[0159] The present invention also provides a system for identifying cell groups based on single-cell transcriptome sequencing data of breast cancer and identifying tumor cell subtypes with EMT or taxane drug resistance characteristics, including:

[0160] Algorithm server and upload, parameter selection and display web page;

[0161] The algorithm server is used to store algorithms and calculate breast cancer single-cell transcriptome sequencing data cell populations and EMT or taxane drug resistance characteristic subtype tumor cell identification, to obtain breast cancer single-cell transcriptome sequencing cell population identification results and EMT or taxane drug resistance characteristic subtype tumor cell identification results;

[0162] The upload, parameter selection and display webpages are integrated on the same website, and are used to upload single-cell transcriptome sequencing data, select the result observation interface and display the calculation results respectively.

[0163] Attached Figure 2 The module and specific process of data uploading and system analysis of the present invention are shown, including:

[0164] S1: The user uploads breast cancer single-cell transcriptome expression matrix data through the upload option of the visualization webpage. These data files are cell gene expression files after the 10X Genomics official analysis software Cell Ranger filters the raw data and aligns genes, including: barcodes.tsv.gz, features.tsv.gz, matrix.mtx.gz, and then executes S2;

[0165] S2: The algorithm server transfers the file to the backend, calls the R script, and then executes S3;

[0166] S3: The computational module in the R script performs cell population identification on the uploaded expression matrix and then executes S4;

[0167] S4: The computational module in the R script determines whether there are tumor cells in the cell cluster. If yes, S5 is executed, otherwise S6 is executed;

[0168] S5: The computational module in the R script calculates the characteristics of tumor cell EMT or taxane-resistant subtype tumor cells, and then executes S6;

[0169] S6: The algorithm server stores the result locally and returns it to the front end, and then executes S7;

[0170] S7: The user selects the result observation interface parameters, and the selectable parameters include: cell group identification pictures and tables, tumor cell EMT or taxane drug resistance score pictures, EMT characteristic tumor cells or taxane drug resistance characteristic tumor cells differentially expressed gene tables, EMT characteristic tumor cells or taxane drug resistance tumor cells pathway enrichment pictures and tables, and then executes S8;

[0171] S8: Display the result file and provide it for users to download.

[0172] Attached Figure 3 The display webpage is shown, including: breast cancer single cell transcriptome sequencing data selection and upload area and result display area ( Figure 3 A in

[0173] Breast cancer single-cell transcriptome sequencing data successfully uploaded interface ( Figure 3 B) in;

[0174] Calculation and result loading process interface ( Figure 3 C in );

[0175] Uploaded breast cancer single-cell sequencing data cell type classification results diagram ( Figure 3 D in

[0176] Uploaded breast cancer single cell EMT characteristic tumor cell enrichment pathway diagram ( Figure 3 E in );

[0177] Uploaded breast cancer single cell taxane drug resistance characteristics tumor cell enrichment pathway map ( Figure 3 F in

[0178] Uploaded breast cancer single cell analysis results table and download interface ( Figure 3 G in.

[0179] The present invention provides a method and system for identifying cell groups and EMT or taxane-resistant characteristic subtype tumor cells from single-cell transcriptome sequencing data of breast cancer. The innovation is mainly reflected in the following three aspects: First, the present invention proposes a new method for identifying cell groups from single-cell sequencing data of breast cancer, which integrates the marker genes of basic cell types of breast cancer tissue and for the first time assigns calculation weights based on the number of occurrences of cell type-specific genes in the literature and constructs a cell group identification function. Second, the present invention can automatically obtain EMT and taxane-resistant subtype tumor cells and their gene expression characteristics, which is helpful to understand the pathogenesis of breast cancer patients and optimize the diagnosis and treatment process of patients. Third, the present invention provides an algorithm server and a visualization webpage that can automatically analyze and present the results of single-cell transcriptome sequencing of breast cancer.

[0180] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for identifying a breast cancer cell population, characterized in that: The following steps are involved: (1) Perform dimensionality reduction and clustering on single-cell sequencing expression matrix data to obtain clustered cell populations; (2) Calculate the correlation between the clustered cell groups obtained in step (1) and each cell type in the breast cancer tissue cell-specific marker data set to determine the cell type corresponding to each clustered cell group.

2. The method for identifying breast cancer cell populations according to claim 1, wherein: The breast cancer tissue cell-specific marker dataset is constructed by the following steps: S1: Obtaining marker genes of basic cells of breast cancer tissue, wherein the basic cells of breast cancer tissue are 15 types, namely, epithelial cells, endothelial cells, fibroblasts, tumor-associated fibroblasts, T cells, Treg cells, B cells, NK cells, NKT cells, macrophages, monocytes, dendritic cells, mast cells, tumor cells and tumor stem cells; S2: sorting the marker genes of each basic cell obtained in step S1, and sorting them from high to low importance according to the ranking of the marker genes in the CellMarker2.0 database, and obtaining the top 10 marker genes of each basic cell; S3: The weight of each marker gene of each basic cell is obtained by calculating the number of times each marker gene of each basic cell appears in the CellMarker2.0 database / the total number of times the top 10 marker genes of this basic cell appear in the CellMarker2.0 database; S4: constructing a breast cancer tissue cell-specific marker dataset based on the basic cells of the breast cancer tissue and the marker genes of each basic cell obtained in step S1, and the weights of each marker gene of each basic cell obtained in step S3.

3. The method for identifying breast cancer cell populations according to claim 2, characterized in that: The step (2) specifically comprises the following steps: S1: finding the top 10 marker genes of the 15 basic cells in the breast cancer tissue cell-specific marker dataset in the clustered cell population, and obtaining the average expression level of each marker gene in each clustered cell population; S2: Normalize the average expression levels of the top 10 marker genes of the 15 basic cell types in each clustered cell group to obtain the expression scores of the top 10 marker genes of each basic cell type in each clustered cell group; S3: For each cell type, the expression scores of the top 10 marker genes in each clustered cell group are multiplied by their weights, and then summed to obtain the classification score of each cell group; S4: The cell type corresponding to the highest classification score of each clustered cell group is marked as the cell type of the respective clustered cell group.

4. The method for identifying breast cancer cell populations according to claim 1, wherein: Before the dimensionality reduction, quality control and standardization processing are also included.

5. A method for identifying characteristics of breast cancer cell metastasis or drug-resistant subtype tumor cells, characterized in that: The following steps are involved: (1) Calculate the relative expression of epithelial-mesenchymal transition marker genes or taxane drug marker genes in breast cancer tumor cell populations; (2) extracting cells with a relative expression of epithelial-mesenchymal transition marker genes greater than 0 and setting them as epithelial-mesenchymal transition marker gene characteristic tumor cells, and extracting cells with a relative expression of epithelial-mesenchymal transition marker genes less than 0 and setting them as non-epithelial-mesenchymal transition marker gene characteristic tumor cells; Cells with taxane resistance expression levels greater than 0 are extracted and set as taxane resistance characteristic tumor cells, and cells with taxane resistance expression levels less than 0 are extracted and set as non-taxane resistance characteristic tumor cells; (3) Separating tumor cells with epithelial-mesenchymal transition and non-epithelial-mesenchymal transition characteristics, and obtaining violin plots of the epithelial-mesenchymal transition scores of the two; separating tumor cells with taxane resistance and non-taxane resistance characteristics, and obtaining violin plots of the taxane resistance scores of the two; (4) Identify differentially expressed genes in tumor cells with epithelial-mesenchymal transition and non-epithelial-mesenchymal transition characteristics, and identify differentially expressed genes in tumor cells with taxane-resistant characteristics and taxane-non-resistant characteristics; (5) Perform KEGG or GO enrichment analysis on several genes with significant differences among the differentially expressed genes obtained in step (4) to obtain enrichment pathways and visualized bubble diagrams.

6. A system for identifying breast cancer cell populations, characterized in that: include: Obtaining a clustered cell population module: The clustered cell population module is used to reduce the dimension and cluster the single-cell sequencing expression matrix data; Cell cluster identification module: The cell cluster identification module is used to calculate the correlation between the clustered cell clusters and each cell type in the breast cancer tissue cell-specific marker data set to determine the cell type corresponding to each clustered cell cluster.

7. A breast cancer cell metastasis or drug-resistant subtype tumor cell feature recognition system, characterized in that: include: Calculation module: The calculation module is used to calculate the relative expression of epithelial-mesenchymal transition marker genes or taxane drug marker genes in breast cancer tumor cell populations; Extraction and setting module: The extraction and setting module is used to extract cells with a relative expression of epithelial-mesenchymal transition marker genes greater than 0 and set them as epithelial-mesenchymal transition marker gene characteristic tumor cells, and extract cells with a relative expression of epithelial-mesenchymal transition marker genes less than 0 and set them as non-epithelial-mesenchymal transition marker gene characteristic tumor cells; Cells with taxane resistance expression levels greater than 0 are extracted and set as taxane resistance characteristic tumor cells, and cells with taxane resistance expression levels less than 0 are extracted and set as non-taxane resistance characteristic tumor cells; Obtaining score violin plot module: The obtaining score violin plot module is used to separate tumor cells with epithelial-mesenchymal transition characteristics and non-epithelial-mesenchymal transition characteristics, obtain violin plots of epithelial-mesenchymal transition scores for both, separate taxane-resistant and non-taxane-resistant tumor cells, and obtain violin plots of taxane-resistant scores for both; Identification module: The identification module is used to identify differentially expressed genes in tumor cells with epithelial-mesenchymal transition and non-epithelial-mesenchymal transition characteristics, and to identify differentially expressed genes in tumor cells with taxane-resistant characteristics and taxol-non-resistant characteristics; Enrichment module: The enrichment module is used to perform KEGG or GO enrichment analysis on several genes with significant differences among the differentially expressed genes obtained, and obtain enrichment pathways and visualized bubble diagrams.

Citation Information

Patent Citations

  • Application of EMT (epithelial to meschymal transition) circulating tumor stem cells in lung cancer proliferation, drug resistance and transfer diseases

    CN107299084A

  • Analysis method based on 10X unicell transcriptome sequencing data

    CN109979538A

  • Application of LYPD1 in diagnosis, treatment, prognosis and recurrence prediction of hepatocellular carcinoma

    CN111876483A

  • Tumor single cell transcriptome sequencing data cell type identification method and system

    CN117037907A

  • Esophageal cancer treatment target screening method based on single cell space transcriptome

    CN118116453A

Cited By

  • Intelligent cell type annotation method based on key marker gene

    CN121148497A