A cell classification method, medium, and device based on single-cell genome expression

Abstract

This invention relates to a cell classification method, medium, and device based on single-cell genome expression. It involves constructing a re-divided BED file based on genome sequence data sets and converting it into a predefined format; establishing a reference between genome sequence data sets to obtain a T2T genome reference; processing the genome sequence data sets and raw real data with the reference information to obtain genome sequence data representations and corresponding cell type labels; inputting the processed genome sequence data representations into a constructed fully connected network model and training it to preset conditions; extracting the genome sequence data representations to be classified, processing them, and inputting them into the trained fully connected network model to obtain the classification results; and implementing the method using the provided medium and device. This invention fully utilizes genome expression information to predict cell types; the re-division of the genome facilitates the extraction of more gene information, improving the accuracy of cell type identification.

Description

Technical Field

[0001] This invention relates to the field of bioinformatics, particularly to information and communication technologies applicable to the processing of genetic or protein-related data in computational molecular biology, and more particularly to a cell classification method, medium, and device based on single-cell genome expression. Background Technology

[0002] Single-cell RNA sequencing (scRNA-seq) is currently one of the main methods for defining cell states at the molecular level. Analysis of scRNA-seq datasets typically begins with dimensionality reduction and clustering. Cell type annotation methods based on scRNA-seq data involve dimensionality reduction and clustering, followed by cell type annotation for each cluster based on the characteristic marker genes expressed by the cell clusters (stromal cells). However, clusters represent groups of cells with relatively similar gene expression profiles, and cells clustered together may belong to the same or similar cell types. Because scRNA-seq datasets may not fully capture different cellular phenomena such as cell transitions, cells may be incorrectly assigned. Furthermore, manual cell type annotation is usually very time-consuming and has limited reproducibility. It requires personnel with specialized knowledge to annotate cell types based on the expression levels of marker genes in each cell cluster and the corresponding markers in the database, which introduces a probability of bias.

[0003] Due to variations in operator skill and sequencing depth across different platforms, scRNA-seq data suffers from batch effects. Furthermore, the genes expressed by each cell at different life stages can differ slightly. Traditional methods, involving manual annotation of cell types, are time-consuming, labor-intensive, and extremely costly. To overcome these limitations, other computational methods for automated cell cluster annotation have emerged. With the rapid development of gene sequencing technology and the rise of artificial intelligence, using deep learning to predict cell types based on gene expression levels has become a key means of achieving cell type annotation. However, current automated cell type annotation methods are based on the same gene segmentation, resulting in many gene sequences in the genome not being fully utilized. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a cell classification method, medium, and device based on single-cell genome expression.

[0005] The concept of this invention lies in the fact that the genome is a complete collection of genetic information of an organism, composed of DNA, including coding and non-coding regions. Current conventional single-cell transcriptome analysis mainly focuses on the expression information of coding regions in the genome, neglecting the potential functional information of a large number of non-coding regions. These neglected regions include, but are not limited to, long non-coding RNAs (lncRNAs), microRNAs (miRNAs), enhancers, promoters, pseudogenes, transposons, and introns, which may play a key role in gene regulation, epigenetic modification, cell fate determination, and the occurrence and development of complex diseases. The invention aims to integrate and utilize the information from these non-coding regions in single-cell transcriptome analysis. This information not only provides a more comprehensive map of gene regulatory networks, but also reveals the functional RNAs outside the coding regions and their roles in different biological processes. Therefore, re-segmenting the genome and making full use of the sequence information in the genome, so that each genome sequence has more gene sequences, can be better used for cell type identification and promote single-cell transcriptome annotation. Here, "having more gene sequences" means discarding the currently agreed length of each genome sequence and redefining the length of each genome sequence. To be precise, it is to increase the length of each genome sequence so that each redefined genome sequence contains more gene sequences (ATGC sequences) than before.

[0006] The technical solution adopted in this invention is a cell classification method based on single-cell genome expression, the method comprising the following steps:

[0007] S1 constructs a re-divided BED file based on the genome sequence data set and converts it into a conventional format file;

[0008] S2 constructs a reference between sets of genomic sequence data to obtain a reference for the T2T genome; here, the T2T genome is a preset (existing) gene division, that is, the division of each gene sequence, which is used to guide the software on how to divide the entire gene sequence into individual genes.

[0009] S3, based on the genome sequence data set and the original real data, is used in conjunction with the reference information of S2, and processed with Cellranger software and the corresponding processing flow to obtain genome sequence data characterization and corresponding cell type labels.

[0010] After processing and characterizing the genomic sequence data from S3, S4 inputs it into the constructed fully connected network model and trains it to preset conditions.

[0011] S5 extracts the genome sequence data to be classified, processes it, and inputs it into the trained fully connected network model to obtain the classification results.

[0012] Preferably, in S1, the BED file based on the genome sequence data set includes the start position, end position, and corresponding gene number of the genome sequence data set.

[0013] Preferably, in S2, the reference data is processed based on the standard data format and CellRanger software.

[0014] Preferably, in S3, after processing, a 10X matrix of genome sequence data is generated, and the corresponding data expression matrix Z is obtained. The rows of Z correspond to the block number of the genome sequence data set, and the columns correspond to each cell. The i-th row and j-th column of Z is the expression level of gene i in cell j.

[0015] Preferably, the cell type labels generated after processing S3 are used to mark the categories, and the correspondence S between each cell number and category is obtained. The training set and test set are then defined from the data representation matrix Z.

[0016] Preferably, in S4, the processing of the genomic sequence data characterization in S3 includes scaling the data in the data expression matrix Z to obtain a new normalized matrix Z with m rows and n columns. new ,

[0017]

[0018] Where min(Z) is the minimum value in matrix Z, max(Z) is the maximum value in matrix Z, and min_val and max_val are the lower and upper limits of the preset scaling range.

[0019] Preferably, the new matrix Z is processed. new To obtain the input Z of the fully connected network model in ,

[0020]

[0021] Among them, Z new i,j For matrix Z new The value of the element in the i-th row and j-th column is given, where i is an integer from 0 to m-1, j is an integer from 0 to n-1, and t is the column number, which is an integer from 0 to n.

[0022] Preferably, in S4, the fully connected network model includes a sequentially connected input layer, several sets of linear processing modules, and an output layer; the linear processing modules are in three sets, and any set of linear processing modules includes a sequentially connected activation function layer, dropout layer, and fully connected layer; the number of nodes in the linear layers of the linear processing modules decreases sequentially.

[0023] A computer-readable storage medium storing a cell classification program based on single-cell genome expression, which, when executed by a processor, implements the aforementioned cell classification method based on single-cell genome expression.

[0024] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, it implements the above-described cell classification method based on single-cell genome expression.

[0025] This invention provides a cell classification method, medium, and device based on single-cell genome expression. The method involves constructing a re-divided BED file based on genome sequence data sets and converting it into a predefined format. A reference between the genome sequence data sets is established to obtain a T2T genome reference. Based on the genome sequence data sets and the original real data, the reference information is processed using CellRanger software and its corresponding processing flow to obtain genome sequence data representations and corresponding cell type labels. The processed genome sequence data representations are then input into a constructed fully connected network model and trained to preset conditions. The genome sequence data representations to be classified are extracted, processed, and then input into the trained fully connected network model to obtain the classification results. The method is implemented using a medium and device.

[0026] The beneficial effects of this invention are that it does not rely on traditional reference genome annotation for cell types, but makes full use of genome expression information to predict cell types; after re-segmenting the genome, each gene block contains a large number of gene sequences, which can make full use of gene blocks that match the cell sequences, thereby facilitating the extraction of more gene information to improve the accuracy of cell type identification. Attached Figure Description

[0027] Figure 1 This is a flowchart of the method of the present invention;

[0028] Figure 2 A flowchart illustrating an embodiment of the present invention;

[0029] Figure 3 The figure shows the accuracy of a single-cell genome expression-based cell classification method in predicting peripheral blood mononuclear cell (PBMC) types. Detailed Implementation

[0030] The present invention will be further described in detail below with reference to embodiments, but the scope of protection of the present invention is not limited thereto.

[0031] This invention relates to a cell classification method based on single-cell genome expression. The following example, using peripheral blood mononuclear cell (PBMC) data, illustrates the method.

[0032] (1) Construct a re-divided BED file based on the genome sequence data set and convert it into a conventional format file;

[0033] First, a BED file for partitioning the T2T human genome sequence was constructed. This file is divided into four columns, representing the genome sequence data set number (23 in total), the start position of genes in the bin (genome sequence data set, the same below), the end position of genes in the bin, and the gene sequence number in the bin. The start position and gene sequence number of the bin for each chromosome start from 1. The size of each bin is 80,000, which is much larger than the existing partition size. When the number of base pairs is insufficient to support the partition size, multiple genes are combined, that is, a bin contains multiple previous genes, thereby obtaining more and richer information. After the BED file is constructed, the bedtogenepred and genepredtogtf tools are used to convert the BED file into a GTF file.

[0034] In practice, the experiment also divided the gene sequence into segments of 5,000, 10,000, 20,000, and 40,000. Different segmentation schemes affect the accuracy of subsequent cell type prediction. In this example, the prediction accuracy was highest when the segmentation was 80,000. However, this data is not unique when dividing different tissues.

[0035] (2) Construct references between genome sequence data sets to obtain a reference for the T2T genome;

[0036] The cellranger mkref pipeline of cellranger is used to construct bin gene block references. The parameters used are "-genome T2T-bin-fasta GCF_009914755.1_T2T-CHM13v2.0_genomic.fna-nthreads 10-genes genomic_chr.gtf", where "-genome T2T-bin" indicates the directory where the cellranger results are saved, "-fasta GCF_009914755.1_T2T-CHM13v2.0_genomic.fna" indicates the fasta file used when constructing the reference, which is the fasta file that comes with the T2T genome, "-nthreads 10" indicates the number of CPU threads used when running the cellranger mkref pipeline, and "-genes genomic_chr.gtf" indicates the gtf file (1) used when constructing the reference;

[0037] The `cellranger mkref pipeline` of `cellranger` is used to construct a T2T genome reference. The parameters used are "-genome T2T-fasta GCF_009914755.1_T2T-CHM13v2.0_genomic.fna-nthreads 10-genes genomic.gtf", where "-genome T2T" indicates the directory where the `cellranger` results are saved, "-fasta GCF_009914755.1_T2T-CHM13v2.0_genomic.fna" indicates the FASTA file used when constructing the reference, which is the FASTA file included in the T2T genome, "-nthreads10" indicates the number of CPU threads used when running the `cellranger mkref pipeline`, and "-genes genomic.gtf" indicates the GTF file included in the T2T genome data.

[0038] (3) Based on the genome sequence data set and the original real data, and with the reference information in (2), the data were processed using Cellranger software and the corresponding processing flow to obtain the genome sequence data characterization and the corresponding cell type label.

[0039] Use the CellRanger Count pipeline to generate bin 10X matrix data. The parameters used are "-id PBMC_20K_bin-localcores 24-localmem 200-expect-cells24000-transcriptome T2T-bin-fastqs PBMC_20K_fastqs", where "-id PBMC_20K_bin" represents the directory where the results are saved, "-localcores 24" represents the number of CPU cores used when running the CellRanger Count pipeline, "-localmem 200" represents the amount of memory used, "-expect-cells 24000" represents the estimated number of cells in the sample, "-transcriptome T2T-bin" represents the directory name of the reference generated in step 2), and "-fastqs PBMC_20K_fastqs" represents the directory name of the FastQ files for the sample.

[0040] The CellRanger Count pipeline is used to generate gene 10X matrix data. The parameters used are "-id PBMC_20K_gene -localcores 24 -localmem 200 -expect-cells 24000 -transcriptome T2T -fastqs PBMC_20K_fastqs", where "-id PBMC_20K_gene" represents the directory where the results are saved, "-localcores 24" represents the number of CPU cores used when running the CellRanger Count pipeline, "-localmem 200" represents the amount of memory used, "-expect-cells 24000" represents the estimated number of cells in the sample, "-transcriptome T2T" represents the directory name of the reference generated in step 3), and "-fastqs PBMC_20K_fastqs" represents the directory name of the FastQ files of the sample.

[0041] The generated bin 10X matrix data was processed using the Seurat standard workflow. During the processing workflow, the function GetAssayData(object=seuart_object,assay='RNA') was used to obtain the bin block gene expression matrix Z of the bin 10X matrix data. The rows of Z are bin block numbers and the columns are cell numbers. The generated gene 10X matrix data was processed using the Seurat standard workflow. Based on the manually selected cell types in the processing results, the correspondence between cell numbers and cell types, S, was obtained. 20% of the cells in the bin block gene expression matrix Z were selected as the test set, and the remaining 80% of the cells were selected as the training set.

[0042] In this invention, the Cellranger software and its corresponding processing flow are used to obtain genomic sequence data characterization, which is the data required for subsequent Seurat processing, namely matrix Z. This matrix represents the gene expression level of cells, with row numbers representing genes and column numbers representing cells. From a single row perspective, it indicates the expression level of the gene represented in that row within each cell; from a single column perspective, it indicates the gene expression level of the cell represented in that column. The specific number in row i and column j represents the expression level of gene i in cell j. Here, "expression level" is equivalent to the prominence of a gene in sequencing. Different genes have different expression levels in different cells, and Seurat's cell clustering and classification are based on the gene expression level of the cells.

[0043] Based on this matrix Z, a specific cell can be identified by the column index. For example, the column name of the 5th column is "AAACCCATCGGTAACT", which represents a unique cell. Similarly, a unique gene can be identified by the row.

[0044] The correspondence S refers to the type of each cell obtained after analysis following the Seurat workflow. For example, if the cell “AAACCCATCGGTAACT” is annotated as a T cell type after analysis, then the corresponding label for the cell “AAACCCATCGGTAACT” in the correspondence S is “T cell”. That is, the cell number is “AAACCCATCGGTAACT” and the cell type is “T cell”. “Cell type label” and “label category” both refer to the same concept, namely “T cell”.

[0045] (4) After characterizing the genome sequence data processed in (3), input it into the constructed fully connected network model and train it to the preset conditions;

[0046] The data in the bin block gene expression matrix Z of the bin10X matrix data are scaled to obtain a new matrix Z with m rows and n columns. new Perform the following transformation:

[0047]

[0048] Where min(Z) is the minimum value in matrix Z, max(Z) is the maximum value in matrix Z, and min_val = 1 and max_va = 1000 are manually specified scaling ranges;

[0049] For the new matrix Z new Data processing is performed to obtain the input Z of the neural network. in The formula is as follows:

[0050]

[0051] Among them, Z new i,j For matrix Z new The value of the element in the i-th row and j-th column, 0 ≤ i < m, 0 ≤ j < n;

[0052] Building a fully connected network model:

[0053] The fully connected network consists of an input layer, three fully connected layers, and an output layer. The input layer accepts input from one X node. The number of nodes in the three fully connected layers are 2000, 1000, and 25, respectively. Each fully connected layer uses a ReLU non-linear activation function and a dropout layer, with the dropout probability set to 0.9. The output layer consists of a softmax non-linear function, and the number of output nodes is the total number of cell types. The network is initialized using a Xavier uniform distribution, and the Adam Optimizer method is used to optimize the network weights. The initial learning rate is set to 0.0001, and during training, the learning rate is gradually reduced with each iteration. The learning rate decreases by 0.95 for each training batch, and the multivariate cross-entropy loss function is used.

[0054] Training model parameters involves using the bin block gene expression matrix Z of the processed bin 10X matrix data. in The features are input into the fully connected network, and after feature extraction and backpropagation, the training parameters are obtained by training a total of 100 times.

[0055] (5) Extract the genome sequence data to be classified, process it, and input it into the trained fully connected network model to obtain the classification result;

[0056] Extract gene expression feature information from the test set, input it into the trained model, and obtain the probability distribution of cell type for each cell, which is used to predict cell type;

[0057] Using peripheral blood mononuclear cell (PBMC) data as an example, the accuracy of cell type label prediction results for this cell dataset was obtained using the above method, such as... Figure 3 As shown, the training accuracy reached 98.95%, and the testing accuracy reached 98.36%.

[0058] The present invention also relates to a computer-readable storage medium storing a cell classification program based on single-cell genome expression, which, when executed by a processor, implements the aforementioned cell classification method based on single-cell genome expression.

[0059] The present invention also relates to a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, it implements the above-described cell classification method based on single-cell genome expression.

[0060] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0061] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0062] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0063] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0064] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0065] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A cell classification method based on single-cell genome expression, characterized in that: The method includes the following steps: S1 constructs a re-divided BED file based on the genome sequence data set and converts it into a conventional format file; S2 constructs inter-group references for genome sequence data to obtain a reference for the T2T genome; S3, based on the genome sequence data set and the original real data, is combined with the reference information of S2, and processed using CellRanger software and the corresponding processing flow to obtain genome sequence data characterization and corresponding cell type labels. S4 processes the genomic sequence data from S3 and then inputs it into the constructed fully connected network model, training it to preset conditions. S5 extracts the genome sequence data to be classified, processes it, and inputs it into the trained fully connected network model to obtain the classification results.

2. The cell classification method based on single-cell genome expression according to claim 1, characterized in that: In S1, the BED file based on the genome sequence data set includes the start position, end position, and corresponding gene number of the genome sequence data set.

3. The cell classification method based on single-cell genome expression according to claim 1, characterized in that: In S2, reference data processing is completed based on standard data formats and CellRanger software.

4. The cell classification method based on single-cell genome expression according to claim 1, characterized in that: In S3, a 10X matrix of genome sequence data is generated after processing, and the corresponding data expression matrix Z is obtained. The rows of Z correspond to the block number of the genome sequence data set, and the columns correspond to each cell.

5. The cell classification method based on single-cell genome expression according to claim 4, characterized in that: After processing S3, the generated cell type labels are labeled to obtain the correspondence S between each cell number and category. The training set and test set are then defined from the data representation matrix Z.

6. The cell classification method based on single-cell genome expression according to claim 4, characterized in that: In S4, the processing of the genomic sequence data characterization from S3 includes scaling the data in the data expression matrix Z to obtain a new normalized matrix with m rows and n columns. , ， Where min(Z) is the minimum value in matrix Z, max(Z) is the maximum value in matrix Z, and min_val and max_val are the lower and upper limits of the preset scaling range.

7. The cell classification method based on single-cell genome expression according to claim 6, characterized in that: Processing new matrices To obtain the input of the fully connected network model , ， in, For matrix The value of the element in the i-th row and j-th column is given, where i is an integer from 0 to m-1, j is an integer from 0 to n-1, and t is the column number.

8. The cell classification method based on single-cell genome expression according to claim 1, characterized in that: In S4, the fully connected network model includes a sequentially connected input layer, several sets of linear processing modules, and an output layer; the linear processing modules are in three sets, and any set of linear processing modules includes a sequentially connected activation function layer, dropout layer, and fully connected layer; the number of nodes in the linear layers of the linear processing modules decreases sequentially.

9. A computer-readable storage medium, characterized in that: It stores a cell classification program based on single-cell genome expression, which, when executed by a processor, implements the cell classification method based on single-cell genome expression as described in any one of claims 1 to 8.

10. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, it implements the cell classification method based on single-cell genome expression as described in any one of claims 1 to 8.

Images