Cell identification method and device based on differential gene screening and electronic equipment

Abstract

The invention provides a cell identification method and device based on differential gene screening and electronic equipment, and relates to the technical field of cell identification, the method comprises the following steps: pre-processing a pre-obtained cell expression matrix to determine cell clusters, then carrying out differential expression analysis on the cells to determine effective cell populations, and then characteristic gene selection is carried out on the effective cell population to realize cell identification, so that the technical problems of low accuracy and stability in the prior art are solved, and the technical effect of improving the accuracy and reliability of cell identification is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of cell identification, and in particular to a cell identification method, device and electronic equipment based on differential gene screening. Background Art

[0002] Single-cell transcriptome sequencing technology (scRNA-seq) has become an important tool for understanding cellular heterogeneity and complex biological systems. In single-cell transcriptome sequencing, the capture of single cells is achieved through a microfluidic system. Specifically, in this process, magnetic beads, single cells, and related enzymes are encapsulated by oil droplets to form an independent single-cell environment, enabling transcriptome sequencing at the single-cell level. However, not only cells are obtained during the capture process, but also a large number of empty packets, that is, oil droplets that do not contain single cells. The expression of these empty packets is not completely empty, because some fragile cells may break during the capture experiment, contaminating the entire system and forming background RNA. In addition, there are many low-quality and unusable residual cells. Therefore, the cell identification (call cell) step is to extract as many meaningful cells as possible from the background.

[0003] At present, commonly used cell identification methods include Cell Ranger, Empty Drops, and force cell. Among them, Cell Ranger is a widely used single-cell transcriptome sequencing data processing software. The cell identification method of Cell Ranger is mainly based on the count of non-repetitive sequences captured by each oil droplet. The entire sequence information captured by each oil droplet is called a barcode, and the reading of each gene is called the UMI number, that is, the unique molecular identifier. In actual projects, these algorithms may fail to capture cells with a relatively low total number of UMIs, or the filtering conditions may be too loose, so that too many background barcodes are included, which will cause trouble for subsequent data analysis. In other words, the existing cell identification technology has technical problems with low accuracy and stability. Summary of the invention

[0004] The object of the present invention is to provide a cell identification method, device and electronic device based on differential gene screening to alleviate the technical problems of low accuracy and stability of cell identification in the prior art.

[0005] In a first aspect, an embodiment of the present invention provides a cell identification method based on differential gene screening, the method comprising: preprocessing a pre-acquired cell expression matrix to determine a plurality of cell clusters; the pre-acquired cell expression matrix is ​​a counting matrix of cell barcodes acquired by a single-cell transcriptome sequencing data processing device; the counting matrix of the cell barcodes is obtained by counting the specific molecular tags of each barcode; performing differentially expressed gene analysis on each of the above cell clusters to determine an effective cell population; and performing characteristic gene selection on the above effective cell population to achieve cell identification.

[0006] In some optional implementations, the above method also includes: obtaining a counting matrix of cell barcodes as a cell expression matrix through a single-cell transcriptome sequencing data processing device; the above cell expression matrix includes specific molecular tags of several barcodes; preprocessing the pre-acquired cell expression matrix, including: determining a filtering threshold through a machine learning model; using the above filtering threshold to filter the above cell expression matrix to remove empty packets and barcodes with specific molecular tags that are not within the above filtering threshold range.

[0007] In some optional implementations, the pre-acquired cell expression matrix is ​​pre-processed, and the process also includes: performing high-variance gene screening, principal component analysis, dimensionality reduction and visualization processing on the filtered cell expression matrix in sequence; the dimensionality reduction and visualization processing include: processing the data after principal component analysis using unified manifold approximation and projection UMAP to generate a single-cell distribution map.

[0008] In some optional implementations, preprocessing the pre-acquired cell expression matrix further includes: performing cluster analysis on the result after dimensionality reduction processing using the Leiden algorithm to generate cell clustering results.

[0009] In some optional implementations, differentially expressed gene analysis is performed on each of the above-mentioned cell clusters to determine a valid cell population, including: performing differentially expressed gene analysis on each of the above-mentioned cell clusters to determine differentially expressed genes; using the Wilcoxon rank sum test to determine the significance of each of the above-mentioned differentially expressed genes; determining whether the cell cluster is a valid cell population based on the above-mentioned differentially expressed genes and the significance of each of the above-mentioned differentially expressed genes; if the number of the differentially expressed genes that meet the valid conditions in the current cell cluster exceeds 100, then the cell cluster is a valid cell population; the valid conditions include: the correction value is less than 0.05, and the difference multiple is greater than 0, and the proportion of cells expressing the current differentially expressed genes in the current cell population exceeds 0.1.

[0010] In some optional implementations, the feature gene selection for the effective cell population includes: processing the effective cell population using a logistic regression model; the logistic regression model includes:

[0011]

[0012] Where P(y=1|X) represents the probability that a sample belongs to the target cluster given feature X; logit(P) is the logistic function, defined as logit(P)=log(1-PP); β0 is the intercept term of the model; β j is the coefficient corresponding to the differentially expressed gene, j ranges from 1 to n; DEG j is the expression value of the jth differentially expressed gene; β n+1 is the coefficient corresponding to total_counts; total_counts is the total count feature value of each cell; β n+2 is the coefficient corresponding to pct_counts_mito; pct_counts_mito is the percentage of mitochondrial gene counts per cell; the goal of the above logistic regression model is to estimate each coefficient β by maximizing the likelihood function of the logistic regression function j .

[0013] In some optional implementations, the feature gene selection for the effective cell population further includes: adding an L1 regularization term to the logistic regression model to achieve feature gene selection; the expression of the L1 regularization term is:

[0014]

[0015] Among them, λ is the regularization strength parameter; |βj| is the absolute value of each coefficient.

[0016] In a second aspect, an embodiment of the present invention provides a cell identification device based on differential gene screening, the device comprising: a preprocessing module, used to preprocess a pre-acquired cell expression matrix to determine a number of cell clusters; the pre-acquired cell expression matrix is ​​a counting matrix of cell barcodes acquired by a single-cell transcriptome sequencing data processing device; the counting matrix of the cell barcodes is obtained by counting the specific molecular tags of each barcode; a differential gene screening module, used to perform differential expression gene analysis on each of the above cell clusters to determine a valid cell population; a feature selection module, used to select feature genes for the above valid cell population to achieve cell identification.

[0017] In a third aspect, an embodiment of the present invention provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, and when the processor executes the computer program, the steps of any one of the methods described in the first aspect are implemented.

[0018] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores machine-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions cause the processor to execute any method described in the first aspect.

[0019] The present invention provides a cell identification method, device and electronic device based on differential gene screening. The method comprises: determining cell clustering by preprocessing a pre-acquired cell expression matrix, then performing differential expression analysis to determine effective cell populations, and then selecting characteristic genes for the effective cell populations to achieve cell identification, thereby solving the technical problems of low accuracy and stability in the prior art and achieving the technical effect of improving the accuracy and reliability of cell identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0021] Figure 1 A schematic diagram of a process for cell identification based on differential gene screening provided in an embodiment of the present invention;

[0022] Figure 2 A schematic diagram of another cell identification method based on differential gene screening provided in an embodiment of the present invention;

[0023] Figure 3 A result analysis diagram of a cell identification experiment using different algorithms on a mouse lung dataset provided by an embodiment of the present invention;

[0024] Figure 4 A result analysis diagram of a cell identification experiment using different algorithms on a whole blood single cell data set provided by an embodiment of the present invention;

[0025] Figure 5 A result analysis diagram of a cell identification experiment using different algorithms on a mouse brain data set provided by an embodiment of the present invention;

[0026] Figure 6 A schematic diagram of the structure of a cell identification device based on differential gene screening provided by an embodiment of the present invention;

[0027] Figure 7 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0029] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. Some embodiments of the present invention are described in detail below in conjunction with the drawings. The following embodiments and features in the embodiments can be combined with each other without conflict.

[0031] Single-cell transcriptome sequencing (scRNA-seq) technology has important applications in cell heterogeneity research, but it faces many challenges in capturing single cells and background noise. During the single-cell capture process, fragile cells rupture and cause background RNA contamination, which makes empty packets (oil droplets without captured cells) show certain expression signals; some effective cells have low UMI numbers due to low capture efficiency or low transcription levels, which makes them easily misidentified as empty packets or background; different experimental conditions and data sets have large variability, and a single filtering standard is difficult to apply to all situations, which may lead to problems of over-filtering or under-filtering; effective screening of differentially expressed genes (DEGs) and combining biological significance for cell identification requires precise statistical methods and threshold selection.

[0032] Currently, commonly used cell identification methods include Cell Ranger, Empty Drops and force cell, which have the following disadvantages in practical applications.

[0033] Disadvantages of the Cell Ranger method: (1) Too strict: The filtering standard of Cell Ranger based on UMI counts is high, resulting in some valid cells with low UMI being misclassified as background, and cells with low total counts cannot be captured; (2) Ignoring biological significance: This method mainly relies on counting data without fully considering biological expression characteristics, and may ignore some important biological signals.

[0034] Disadvantages of the EmptyDrops method: (1) Too loose: EmptyDrops uses a statistical model to distinguish between empty cells and cells, but in samples with high background RNA, this method may mistake some background RNA for valid cells, resulting in the inclusion of a large number of background cells. (2) Dependence on the statistical model: It relies on a preset background RNA model and cannot adapt to the variability of all data sets.

[0035] Disadvantages of the force cell method: (1) Manually limit the number of cells: The number of cells to be captured needs to be manually set, and it is difficult to find a perfect threshold in this process. (2) Threshold problem: If the threshold is set too low, background cells are likely to be included; if the threshold is set too high, valid cells with low UMI may be missed.

[0036] Based on this, the embodiments of the present invention provide a cell identification method, device and electronic device based on differential gene screening to solve the above problems existing in the prior art.

[0037] To facilitate understanding of this embodiment, firstly, a cell identification method based on differential gene screening disclosed in an embodiment of the present invention is described in detail, see Figure 1 The flowchart of a cell identification method based on differential gene screening is shown, which can be executed by an electronic device and mainly includes the following steps S110 to S130:

[0038] S110: preprocessing the pre-acquired cell expression matrix to determine a number of cell clusters;

[0039] The pre-acquired cell expression matrix is ​​a count matrix of cell barcodes acquired by a single-cell transcriptome sequencing data processing device; the count matrix of cell barcodes is obtained by counting specific molecular tags of each barcode;

[0040] S120: Perform differentially expressed gene analysis on each cell cluster to determine the effective cell population;

[0041] S130: Select characteristic genes of effective cell populations to achieve cell identification.

[0042] In one embodiment, the method further includes:

[0043] A count matrix of cell barcodes is obtained as a cell expression matrix through a single-cell transcriptome sequencing data processing device; the cell expression matrix includes specific molecular tags of several barcodes;

[0044] The step of preprocessing the pre-acquired cell expression matrix includes:

[0045] (S11) determining a filtering threshold through a machine learning model;

[0046] (S12) Filtering the cell expression matrix using a filtering threshold to remove empty packets and barcodes of specific molecular tags that are not within the filtering threshold range.

[0047] In one embodiment, the step of preprocessing the pre-acquired cell expression matrix further includes:

[0048] (S13) performing high-variability gene screening, principal component analysis, dimensionality reduction, and visualization processing on the filtered cell expression matrix in sequence;

[0049] Among them, dimensionality reduction and visualization processing include: using unified manifold approximation and projection UMAP to process the data after principal component analysis to generate a single-cell distribution map.

[0050] In one embodiment, the step of preprocessing the pre-acquired cell expression matrix further includes:

[0051] (S14) Perform cluster analysis on the results after dimensionality reduction using the Leiden algorithm to generate cell clustering results.

[0052] In one embodiment, the above step S120 performs differentially expressed gene analysis on each cell cluster to determine an effective cell population, including:

[0053] (S21) performing differentially expressed gene analysis on each cell cluster to determine the differentially expressed genes;

[0054] (S22) The Wilcoxon rank sum test was used to determine the significance of each differentially expressed gene;

[0055] (S23) determining whether the cell cluster is a valid cell population based on the differentially expressed genes and the significance of each differentially expressed gene;

[0056] (S24) If the number of differentially expressed genes that meet the valid conditions in the current cell cluster exceeds 100, the cell cluster is a valid cell population; the valid conditions include: the correction value is less than 0.05, the difference multiple is greater than 0, and the proportion of cells expressing the current differentially expressed gene in the current cell population exceeds 0.1.

[0057] In one embodiment, the above step S130 performs characteristic gene selection on the effective cell population, including:

[0058] (S31) Processing the effective cell population using a logistic regression model; the logistic regression model includes:

[0059]

[0060]

[0061] Where P(y=1|X) represents the probability that a sample belongs to the target cluster given feature X; logit(P) is the logistic function, defined as logit(P)=log(1-PP); β0 is the intercept term of the model; βj is the coefficient corresponding to differentially expressed genes (DEGs), j ranges from 1 to n; DEG j is the expression value of the jth differentially expressed gene; β n+1 is the coefficient corresponding to total_counts; total_counts is the total count feature value of each cell; β n+2 is the coefficient corresponding to pct_counts_mito; pct_counts_mito is the percentage of mitochondrial gene counts per cell;

[0062] The goal of the logistic regression model is to estimate each coefficient β by maximizing the likelihood function of the logistic regression function. j .

[0063] In one embodiment, the above step S130 of selecting characteristic genes for the effective cell population further includes:

[0064] (S32) Adding L1 regularization term to the logistic regression model to achieve feature gene selection; the expression of L1 regularization term is:

[0065]

[0066] Among them, λ is the regularization strength parameter; |βj| is the absolute value of each coefficient.

[0067] The above-mentioned cell identification method based on differential gene screening provided in this embodiment combines differential gene screening and logistic regression algorithm, aiming to more accurately extract meaningful cells from the background, so as to improve the accuracy of cell identification, effectively identify effective cells with low UMI, and avoid excessive filtering; it shows higher flexibility and accuracy in processing background RNA and low UMI cells, and can effectively solve the problems existing in the prior art.

[0068] As a specific example, an embodiment of the present invention provides a cell identification method based on differential gene screening, which combines differential gene screening and logistic regression algorithm for cell identification and has advantages over existing methods.

[0069] The method can be performed by an electronic device, which includes a display screen and a processor, wherein the display screen is used to present a single-cell transcriptome sequencing system, which includes a single-cell transcriptome sequencing data processing software cell ranger, an input unit, a processing unit, and an output unit. Figure 2 As shown, in combination with the above system, the method includes:

[0070] (1) Read the unfiltered expression matrix file obtained by cellranget software through the input unit;

[0071] (2) Perform preliminary filtering in the processing unit to obtain 20UMI barcodes;

[0072] (3) After screening highly variable genes, PCA and UMAP dimensionality reduction, and clustering by clustering algorithms in the processing units, the differences are found based on the clustering results;

[0073] (4) Filter the cell populations with a significant p-value of more than 100 genes in the processing unit and directly identify them as cells, and continue to filter the remaining cell populations;

[0074] (5) Using a logistic regression algorithm in the processing unit, and considering the number of cell UMIs, mitochondrial ratios, and differential genes in the cell population, we try to find genes with high specificity;

[0075] (6) In the output result unit, the cell population that passes all the filters is identified as cells, and the remaining cell population is identified as background.

[0076] The specific steps are as follows:

[0077] 1. Load the counts matrix of all barcodes: Read the initial input data through the input unit, that is, the unfiltered expression matrix file obtained by the cellranger software, including the UMI counts of all barcodes.

[0078] 2. Filter out empty packets and barcodes with rare UMI counts: Use 20UMI as the threshold to filter out obvious empty packets and barcodes with very few UMI counts.

[0079] 3. Run the HVGs→PCA→UMAP→Leiden pipeline for clustering: screen highly variable genes (HVGs), perform dimensionality reduction and visualization through PCA (principal component analysis) and UMAP, and use the Leiden algorithm for clustering.

[0080] 4. Calculation of differentially expressed genes (DEGs) by Wilcoxon rank-sum test: Differentially expressed genes were analyzed for each cluster, and the significance of each gene was calculated using the Wilcoxon rank-sum test.

[0081] 5. Identification of cell clusters: For clusters with more than 100 differentially expressed genes (p_adj < 0.05, logfoldchange > 0, pct_nz_group > 0.1, that is, the corrected p value is less than 0.05, the difference fold is greater than 0, and the proportion of cells expressing the gene in the cell group is more than 0.1), they are considered to be valid cell populations.

[0082] 6. Use logistic regression and L1 regularization to identify characteristic genes: Apply a logistic regression model to each cell cluster and use L1 regularization to select characteristic genes. The specific formula is as follows:

[0083]

[0084] Where P(y=1|X) represents the probability that a sample belongs to the target cluster given feature X. Logit(P) is the logistic function, defined as logit(P)=log(1-PP). β0 is the intercept term of the model. βj is the coefficient corresponding to the differentially expressed genes (DEGs), j ranges from 1 to n.

[0085] DEGj is the expression value of the jth differentially expressed gene.

[0086] βn+1 is the coefficient corresponding to total_counts.

[0087] total_counts is the total count feature value of each cell.

[0088] βn+2 is the coefficient corresponding to pct_counts_mito.

[0089] pct_counts_mito is the percentage of mitochondrial gene counts per cell.

[0090] The goal of this model is to estimate each coefficient βj by maximizing the likelihood function of the logistic regression function. To prevent overfitting, we apply L1 regularization in the model.

[0091] L1 Regularization:

[0092] In the logistic regression model, the addition of L1 regularization can prevent the model from overfitting and enable feature selection. The expression of L1 regularization is:

[0093]

[0094] Where: λ is the regularization strength parameter (i.e. the inverse of C in the code). |βj| is the absolute value of each coefficient.

[0095] By optimizing this regularized likelihood function, we can obtain the optimal coefficient estimates βj, which can not only reflect the relationship between the features and the target variables, but also select the most influential features through L1 regularization.

[0096] The above method combines differential gene screening based on biological significance and logistic regression algorithm, not only considering the total number of counts, but also considering the biological expression characteristics, thus achieving more accurate cell identification. In addition, this method performs better when capturing cells with a low total number of counts, while avoiding the problem of overly loose filtering conditions and effectively reducing the interference of background barcodes.

[0097] Through this method, meaningful cells can be extracted from the background more accurately, providing a higher quality data foundation for subsequent data analysis.

[0098] Based on a cell identification method based on differential gene screening provided in an embodiment of the present invention, a cell identification algorithm applied to single-cell sequencing data was developed, and the algorithm was implemented using the Python programming language, achieving automated, accurate and rapid background removal to obtain effective cell identification results.

[0099] The following is an analysis of the results of cell identification experiments using different algorithms on multiple data sets:

[0100] 1. Mouse lung dataset, such as Figure 3 As shown:

[0101] Cell Ranger: This method is too strict, and many valid cells with low UMI, such as cells in cluster 10, are not captured.

[0102] EmptyDrops: This method is too lenient and includes a large number of background cells, such as cluster 1-9 in the figure, resulting in a very low median number of genes, only 179.

[0103] like Figure 3 As shown in the upper left figure, after dimensionality reduction clustering of all cells including the background, background cells with low UMI counts are always clustered together to form a circular shape, while target cells with higher UMI counts are located at the edge of these background cells or far away from them. Our goal is to capture these cells with high UMI counts that can be distinguished from the background.

[0104] When using the existing algorithm such as emptydrop, as shown in the red circle in the UMAP graph in the lower right corner of the upper left figure, the algorithm will capture a large number of background cells, resulting in too many cells in the end, and the median gene counts and UMI counts are very low (see Figure 3 The table below shows that this will not only increase the resource consumption of subsequent analysis, but also greatly increase the background noise, interfering with the downstream analysis results. The algorithm provided by the embodiment of the present invention and the cellranger algorithm have similar performance in this data set, and both can effectively remove the background and retain high-quality cells.

[0105] 2. Whole blood single cell dataset, such as Figure 4 As shown:

[0106] Cell Ranger: It also exhibits strict filtering criteria, resulting in some valid cells in clusters 3, 7, and 5 not being captured.

[0107] EmptyDrops: performs moderately in this dataset, but is too loose in other datasets, showing instability between datasets.

[0108] Force cell: The manually set cell number threshold makes it difficult to balance the identification of valid cells and background cells, and the results also include some background cells.

[0109] like Figure 4 As shown in the UMAP diagram on the upper left, the traditional cellranger algorithm cannot effectively capture cells with low UMI counts that are significantly different from the background. This can be seen from the red circle in the figure. These cells are far away from the background cells and close to them at the same time. However, the cellranger algorithm only identifies a small number of cells as the final cells, which will cause the proportion of the cell population in the downstream analysis to change, and may miss key biological information. The cellranger software provides a possible solution, that is, manually specifying the cells with the most UMI counts in the first 8,000 as the final cells, but this method will not only introduce human errors, but also, as can be seen from the yellow circle in the figure, may introduce some low-quality background cells. In the human whole blood single cell data, the algorithm and the emptydrop algorithm provided in the embodiment of the present invention can both better capture cells with low UMI counts at the red circle position but significantly different from the background.

[0110] 3. Mouse brain dataset, such as Figure 5 As shown:

[0111] Cell Ranger: This method is too strict. Only barcodes with very high counts are identified as cells. Its median number of genes reaches 4319.

[0112] EmptyDrops: This method is too lenient and includes a large number of background cells, such as cluster 1-9 in the figure. As a result, the final number of cells is too large, which is more than ten times that of the cellranger method.

[0113] like Figure 5 The UMAP graph in Figure 1 clearly shows that the emptydrop algorithm is too lenient. It captures far more cells than the other two algorithms and includes a large number of background cells. Figure 5 It can also be seen from the curve graph below (in the red circle) that some of the cells captured by the algorithm are at the very end of the curve, which are all background cells with extremely low UMI counts. The algorithm and cellranger algorithm provided by the embodiment of the present invention can better remove the background. In this data set, the algorithm provided by the embodiment of the present invention is more relaxed than the cellranger algorithm, but more stringent than emptydrop. It is a compromise algorithm that can retain cells with low UMI counts while better removing the background, which also reflects the advantages of the algorithm.

[0114] From the above three data sets, we can see that although emptydrop can capture cells with low UMI counts (whole blood single cell data set), it may introduce a large number of background cells (mouse lung, mouse brain data set) due to its overly loose screening method. The cellranger algorithm is too strict, and it easily ignores some cells with low UMI counts but obviously different from the background (whole blood single cell data set). The algorithm provided by the embodiment of the present invention has stable and good performance in the above three data sets, that is, it can retain cells with low UMI counts but different from the background, and can remove most of the background cells (the three data sets are clustered together in the UMAP map to form an irregular circular area).

[0115] Compared with the existing technology, the above method has significant advantages in the following aspects:

[0116] 1. Comprehensive consideration of biological characteristics: Existing technologies such as Cell Ranger and EmptyDrops mainly rely on UMI counts for cell identification, while the cell identification method provided in the embodiment of the present invention combines the biological characteristics of differentially expressed genes (DEGs) and screens significantly differentially expressed genes through the Wilcoxon rank sum test, making cell identification more biologically meaningful.

[0117] 2. Improve the capture rate of low UMI cells: The Cell Ranger method is too strict and easily misses low UMI cells; the EmptyDrops method is too loose and easily includes background cells. The cell identification method provided in the embodiment of the present invention combines the logistic regression model and L1 regularization to not only effectively identify low UMI cells, but also avoid over-filtering and background interference.

[0118] 3. Automated threshold selection: Existing methods such as force cell require manual threshold setting, which makes it difficult to find a perfect balance. The cell identification method provided by the embodiment of the present invention uses statistical analysis and machine learning models to automatically screen appropriate thresholds, reduce the impact of human factors, and adapt to the variability of different data sets.

[0119] 4. Accurately identify effective cell populations: The cell identification method provided in the embodiment of the present invention improves the accuracy and reliability of cell identification by identifying characteristic genes through multi-step screening and logistic regression based on cluster analysis, and provides a higher quality cell population for subsequent data analysis.

[0120] In summary, the method provided in the embodiment of the present invention performs well in improving the accuracy of cell identification, processing low UMI cells and reducing background interference, overcomes the shortcomings of the prior art, and provides stronger support for single-cell transcriptome sequencing data analysis.

[0121] In addition, the present invention also provides a cell identification device based on differential gene screening, see Figure 6 As shown, the device comprises:

[0122] The preprocessing module 310 is used to preprocess the pre-acquired cell expression matrix to determine a number of cell clusters; the pre-acquired cell expression matrix is ​​a count matrix of cell barcodes obtained by a single-cell transcriptome sequencing data processing device; the count matrix of cell barcodes is obtained by counting specific molecular tags of each barcode;

[0123] The differential gene screening module 320 is used to perform differentially expressed gene analysis on each cell cluster to determine the effective cell population;

[0124] The feature selection module 330 is used to select feature genes for effective cell populations to achieve cell identification.

[0125] The cell identification device based on differential gene screening provided in the embodiment of the present application can be specific hardware on the device or software or firmware installed on the device, etc. The device provided in the embodiment of the present application, its implementation principle and the technical effect produced are the same as those in the aforementioned method embodiment. For brief description, the device embodiment part is not mentioned, and the corresponding content in the aforementioned method embodiment can be referred to. Technical personnel in the relevant field can clearly understand that for the convenience and simplicity of description, the specific working process of the system, device and unit described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. The cell identification device based on differential gene screening provided in the embodiment of the present application has the same technical features as the cell identification method based on differential gene screening provided in the above embodiment, so it can also solve the same technical problems and achieve the same technical effects.

[0126] An embodiment of the present application further provides an electronic device. Specifically, the electronic device includes a processor and a storage device. The storage device stores a computer program, and when the computer program is executed by the processor, it executes the method described in any one of the above-mentioned embodiments.

[0127] Figure 7 A structural schematic diagram of an electronic device provided in an embodiment of the present application, the electronic device 400 includes: a processor 40, a memory 41, a bus 42 and a communication interface 43, wherein the processor 40, the communication interface 43 and the memory 41 are connected via the bus 42; the processor 40 is used to execute an executable module stored in the memory 41, such as a computer program.

[0128] The memory 41 may include a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk memory. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 43 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used.

[0129] The bus 42 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 7 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0130] Among them, the memory 41 is used to store programs, and the processor 40 executes the program after receiving the execution instruction. The method executed by the process definition device disclosed in any embodiment of the above-mentioned embodiment of the present invention can be applied to the processor 40 or implemented by the processor 40.

[0131] The processor 40 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 40. The above processor 40 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in the embodiments of the present invention can be directly embodied as a hardware decoding processor to execute, or can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 41, and the processor 40 reads the information in the memory 41 and completes the steps of the above method in combination with its hardware.

[0132] Corresponding to the above method, an embodiment of the present application also provides a computer-readable storage medium, which stores machine-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to execute the steps of the above method.

[0133] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, 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 mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0134] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0135] In addition, each functional unit in the embodiments provided in the present application 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.

[0136] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, electronic device, or network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0137] It should be noted that similar numbers and letters represent similar items in the accompanying drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. In addition, the terms "first", "second", "third", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0138] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A cell identification method based on differential gene screening, characterized in that: The method comprises: Preprocessing a pre-acquired cell expression matrix to determine a number of cell clusters; the pre-acquired cell expression matrix is ​​a cell barcode counting matrix acquired by a single-cell transcriptome sequencing data processing device; the cell barcode counting matrix is ​​obtained by counting specific molecular tags of each barcode; Performing differentially expressed gene analysis on each of the cell clusters to determine an effective cell population; The effective cell population is subjected to characteristic gene selection to achieve cell identification.

2. The cell identification method based on differential gene screening according to claim 1, characterized in that: The method further comprises: Obtaining a cell barcode count matrix as a cell expression matrix through a single-cell transcriptome sequencing data processing device; the cell expression matrix includes specific molecular tags of several barcodes; Preprocess the previously acquired cell expression matrix, including: Determine the filtering threshold through machine learning models; The cell expression matrix is ​​filtered using the filtering threshold to remove empty packets and barcodes of specific molecular tags that are not within the filtering threshold range.

3. The cell identification method based on differential gene screening according to claim 2, characterized in that: Preprocessing of the previously acquired cell expression matrix also includes: The filtered cell expression matrix is ​​sequentially subjected to high-variance gene screening, principal component analysis, dimensionality reduction and visualization processing; the dimensionality reduction and visualization processing include: using unified manifold approximation and projection UMAP to process the data after principal component analysis to generate a single cell distribution map.

4. The cell identification method based on differential gene screening according to claim 3, characterized in that: The preprocessing of the pre-acquired cell expression matrix also includes: using the Leiden algorithm to perform cluster analysis on the result after the dimension reduction processing to generate a cell clustering result.

5. The cell identification method based on differential gene screening according to claim 1, characterized in that: Perform differentially expressed gene analysis on each of the cell clusters to determine effective cell populations, including: Performing differentially expressed gene analysis on each of the cell clusters to determine the differentially expressed genes; The Wilcoxon rank sum test was used to determine the significance of each differentially expressed gene; Determining whether the cell cluster is a valid cell population according to the differentially expressed genes and the significance of each of the differentially expressed genes; If the number of the differentially expressed genes that meet the effective conditions in the current cell cluster exceeds 100, the cell cluster is a valid cell population; the effective conditions include: the correction value is less than 0.05, the difference multiple is greater than 0, and the proportion of cells expressing the current differentially expressed genes in the current cell population exceeds 0.

1.

6. The cell identification method based on differential gene screening according to claim 1, characterized in that: Selecting characteristic genes from the effective cell population includes: The effective cell population is processed using a logistic regression model; the logistic regression model includes: Where P(y=1|X) represents the probability that a sample belongs to the target cluster given feature X; logit(P) is the logistic function, defined as logit(P)=log(1-PP); β0 is the intercept term of the model; βj is the coefficient corresponding to the differentially expressed gene, j ranges from 1 to n; DEG j is the expression value of the jth differentially expressed gene; β n+1 is the coefficient corresponding to total_counts; total_counts is the total count feature value of each cell; β n+2 is the coefficient corresponding to pct_counts_mito; pct_counts_mito is the percentage of mitochondrial gene counts per cell; The goal of the logistic regression model is to estimate each coefficient β by maximizing the likelihood function of the logistic regression function. j .

7. The cell identification method based on differential gene screening according to claim 6, characterized in that: Selecting characteristic genes from the effective cell population also includes: An L1 regularization term is added to the logistic regression model to achieve feature gene selection; the expression of the L1 regularization term is: Among them, λ is the regularization strength parameter; |βj| is the absolute value of each coefficient.

8. A cell identification device based on differential gene screening, characterized in that: include: A preprocessing module, used to preprocess a pre-acquired cell expression matrix to determine a number of cell clusters; the pre-acquired cell expression matrix is ​​a cell barcode counting matrix acquired by a single-cell transcriptome sequencing data processing device; the cell barcode counting matrix is ​​obtained by counting specific molecular tags of each barcode; A differential gene screening module is used to perform differentially expressed gene analysis on each of the cell clusters to determine the effective cell population; The feature selection module is used to select feature genes for the effective cell population to achieve cell identification.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores machine-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions cause the processor to execute the method according to any one of claims 1 to 7.

Images