A method for annotating cells from non-model species using bidirectional homology alignment

Abstract

The invention discloses a non-model species cell annotation method of bidirectional homology comparison, belongs to the field of cell annotation technology, and mainly comprises the following steps: bidirectional homology comparison is performed on the genome of the non-model species and the genome of the model species, and the gene pairs that are paired consistently in both directions are retained; in the marker gene database of the model species, the genes of the non-model species in the bidirectional homology comparison gene pairs are replaced with the genes homologous to the non-model species in the bidirectional homology comparison gene pairs, so as to obtain the expression matrix of the non-model species; the gene data of the expression matrix of the non-model species is called by software for annotation, so as to obtain the annotation results of the cell types of the non-model species. By adopting the non-model species cell annotation method of bidirectional homology comparison, the genes of the non-model species can be accurately homologously compared, the genes homologous to the model species can be obtained, and the cell annotation of the non-model species can be realized by using the marker gene database of the model species.

Description

Technical Field

[0001] The invention belongs to the technical field of cell annotation, and in particular relates to a non-model species cell annotation method for bidirectional homology alignment. Background Art

[0002] The cell annotation analysis of single-cell transcriptomes is highly dependent on the analysis content of the database. The former is an important basis for giving biological meaning to the data. Single-cell transcriptomes attribute reads to their source cells by adding identification tags to the mRNA of each cell, thereby obtaining the gene expression abundance of single cells. However, in this process, we can only obtain the gene expression level of the cell, but we cannot know what cell type the cell belongs to. In order to break this isolation, a common method is to give cells biologically meaningful cell type labels based on the expression of cell type-related marker genes in the cells, so as to discuss the molecular characteristics of the samples based on cell types in subsequent analysis. However, non-model species lack systematic marker gene research, and marker genes are difficult to obtain, resulting in a lack of research on the expression of cell type-related marker genes in cells, which has become the main obstacle to cell annotation in non-model species.

[0003] Marker genes are important indicators for cell type identification and have been shown to be conserved in many articles. For example, the journal Cell reported the homology of retinal marker genes between marmosets and humans and mice (Peng YR, Shekhar K, Yan W, et al. Molecular classification and comparative taxonomics of foveal and peripheral cells in primate retina[J]. Cell, 2019, 176(5): 1222-1237.e22.). It also reported the homology of hepatocyte marker genes in humans, mice, macaques, pigs, hamsters, chickens, and zebrafish (Guilliams M, Bonnardel J, Haest B, et al. Spatial proteogenomics reveals distinct and evolutionarily conserved hepatic macrophages). niches[J].Cell,2022,185(2):379-396.e38.), and Science magazine reported the homology of marker genes of the claustrum in humans, mice, turtles, and bearded dragons (Norimoto H,Fenk LA,Li HH,et al.A claustrum in reptiles and its role inslow-wave sleep[J].Science,2020,578(7795):413-418.).

[0004] The above reports suggest that through homologous alignment and reference to mature marker gene databases of model species, we can get rid of the current lack of systematic marker gene research on non-model species and rely on information related to the distribution of cell type marker genes in single-cell data, which is conducive to achieving a breakthrough from 0 to 1 in single-cell research on non-model species. However, the homologous alignment methods currently used are all one-way alignments, that is, the genes of the species to be analyzed are aligned to the genes of model species (such as mice). This alignment method is prone to the situation where one gene has multiple homologous genes or false positive homologous genes, which interferes with the data analysis results and thus affects the correctness of the results. Summary of the invention

[0005] In order to overcome the above-mentioned deficiencies in the prior art, the purpose of the present invention is to provide a method for annotating cells of non-model species by bidirectional homologous alignment, which can accurately perform homologous alignment on the genes of non-model species, obtain genes homologous to model species, and use genes homologous to the model species database to complete correlation analysis, thereby realizing cell annotation of non-model species.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] The present invention provides a non-model species cell annotation method for bidirectional homology alignment, comprising the following steps:

[0008] S1. Perform single-cell transcriptome sequencing on non-model species to obtain single-cell transcriptome sequencing data;

[0009] S2, aligning the genome of the non-model species with the genome of the model species, and retaining the only aligned gene pairs;

[0010] S3, aligning the genome of the model species with the genome of the non-model species, and retaining the only aligned gene pairs;

[0011] S4, comparing the uniquely aligned gene pairs retained in S2 with the uniquely aligned gene pairs retained in S3, retaining the same gene pairs as the gene pairs for bidirectional homologous alignment;

[0012] S5. In the marker gene database of the model species, the genes of the non-model species in the gene pairs of the bidirectional homologous alignment are replaced by the genes of the non-model species in the gene pairs of the bidirectional homologous alignment to obtain the expression matrix of the non-model species;

[0013] S6. The gene data of the expression matrix of the non-model species is called by software for annotation to obtain annotation results of the cell type of the non-model species.

[0014] Furthermore, in S2 and S3, the genome of the model species and the genome of the non-model species are compared using the blast method.

[0015] Furthermore, in S5, the marker gene database of the model species includes an expression matrix of the model species. In the expression matrix of the model species, the gene ID of the model species in the gene pair of bidirectional homologous alignment is replaced with the gene homologous to the model species in the gene pair of bidirectional homologous alignment to obtain the expression matrix of the non-model species.

[0016] Furthermore, the singleR software is used to call the gene data of the expression matrix of the non-model species for annotation, so as to obtain the annotation results of the cell type of the non-model species.

[0017] Furthermore, after obtaining the annotation results of the non-model species cell type, using the ligand-receptor database of the model species, according to the obtained annotation results of the non-model species cell type, cell communication analysis can be performed on the non-model species cells based on the annotated non-model species cell type.

[0018] Furthermore, after obtaining the annotation results of the non-model species cell types, the annotation results of the non-model species cell types are counted according to the cell types, and the distribution of the non-model species cell types is visualized through a table.

[0019] Furthermore, the marker gene database of the model species includes a marker gene database of humans and a marker gene database of mice.

[0020] Compared with the prior art, the non-model species cell annotation method for bidirectional homology alignment provided by the present invention has the following beneficial effects:

[0021] It is possible to accurately perform homologous comparisons on the genes of non-model species and obtain genes homologous to model species, thus ensuring the accuracy of subsequent data analysis results. Based on the genes homologous to model species, the software calls the gene data of the marker gene database of model species for annotation, thereby realizing the annotation of cell types of non-model species. On the one hand, it improves the cell type analysis of non-model species, thereby establishing a marker gene system belonging to the cells of non-model species themselves. On the other hand, it applies the ligand-receptor database of model species to the cell communication analysis of non-model species, making cell communication analysis of non-model species possible and achieving a breakthrough from 0 to 1. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0023] Figure 1 It is a schematic diagram of the process of the non-model species cell annotation method for bidirectional homology alignment provided in Example 1.

[0024] Figure 2 This is the manually annotated t-SNE diagram of rat cells provided in Example 2.

[0025] Figure 3 This is the t-SNE diagram of the rat genome blast bidirectional homology alignment to the mouse provided in Example 2.

[0026] Figure 4 This is the t-SNE diagram of the rat genome ensembl bidirectional homology alignment to the mouse provided in Example 2.

[0027] Figure 5This is the t-SNE diagram of the bidirectional homology alignment of the rat genome blast to human provided in Example 3.

[0028] Figure 6 This is the t-SNE diagram of the rat genome ensembl bidirectional homology alignment to human provided in Example 3.

[0029] Figure 7 The UMAP map is obtained by annotating cell types using the non-model species cell annotation method of bidirectional homology alignment provided by the present invention based on the data provided in the article.

[0030] Figure 8 This is the original UMAP image of rats compared to humans obtained in the article.

[0031] Fig. 9 This is the UMAP map of rat versus human obtained by the non-model species cell annotation method using bidirectional homology alignment in Example 5.

[0032] Fig.10 This is the UMAP map of rat versus mouse obtained by the non-model species cell annotation method using bidirectional homologous alignment in Example 6.

[0033] Fig.11 It is a Sankey diagram reflecting the relationship between the cell type annotation results obtained in Example 6 and the article. DETAILED DESCRIPTION

[0034] In order to make the technical problems, technical solutions and beneficial effects to be solved by this application clearer, the technical solutions of this application will be described clearly and completely in combination with the embodiments below. It should be understood that the specific embodiments described here are only used to explain this application and are not used to limit this application.

[0035] Example 1

[0036] Please see attached Figure 1 The present embodiment provides a non-model species cell annotation method for bidirectional homology alignment, comprising the following steps:

[0037] S1. Perform single-cell transcriptome sequencing on non-model species to obtain single-cell transcriptome sequencing data;

[0038] S2, aligning the genome of the non-model species with the genome of the model species, and retaining the only aligned gene pairs;

[0039] S3, aligning the genome of the model species with the genome of the non-model species, and retaining the only aligned gene pairs;

[0040] S4, comparing the gene pairs retained in the unique alignment in S2 with the gene pairs retained in S3, retaining the same gene pairs as the gene pairs for bidirectional homologous alignment;

[0041] Understandably, the gene pairs obtained by bidirectional homologous alignment of the genomes of model species and non-model species are very accurate, and the existing homologous alignment methods that use unidirectional alignments are prone to produce multiple homologous genes for one gene or false positive homologous genes, thereby ensuring the accuracy of subsequent data analysis results.

[0042] Specifically, in one implementation of the present invention, when the non-model species has an ensembl ID number, the ensembl method is used to perform a bidirectional homology comparison between the genome of the model species and the genome of the non-model species.

[0043] In one implementation of the present invention, the blast method is directly used to perform bidirectional homology comparison on the genome of the model species and the genome of the non-model species.

[0044] S5. In the marker gene database of the model species, the genes of the non-model species in the gene pairs of the bidirectional homologous alignment are substituted for the genes homologous to the non-model species in the gene pairs of the bidirectional homologous alignment to obtain the expression matrix of the non-model species;

[0045] Specifically, the marker gene database of the model species includes the expression matrix of the model species. In the expression matrix of the model species, the gene ID of the model species in the gene pair of the bidirectional homologous alignment is replaced with the gene homologous to the model species in the gene pair of the bidirectional homologous alignment to obtain the expression matrix of the non-model species.

[0046] S6. The gene data of the expression matrix of the non-model species is called by software for annotation to obtain the annotation results of the cell type of the non-model species.

[0047] Specifically, the singleR software is used to call the gene data of the expression matrix of the non-model species for annotation, and the annotation results of the cell types of the non-model species are obtained.

[0048] It should be noted that after obtaining the annotation results of the non-model species cell types, using the ligand-receptor database of the model species, according to the obtained annotation results of the non-model species cell types, cell communication analysis between non-model species cells can be performed based on the annotated non-model species cell types.

[0049] In addition, after obtaining the annotation results of the non-model species cell types, the annotation results of the non-model species cell types can also be counted according to the cell types, and the distribution of the non-model species cell types can be visualized through a table.

[0050] The method of the present invention can accurately perform homologous comparison on genes of non-model species, obtain genes homologous to model species, and ensure the accuracy of subsequent data analysis results. According to the obtained genes homologous to model species, the gene data of the marker gene database of model species are called by software for annotation, so as to realize the annotation of cell types of non-model species. On the one hand, the cell type analysis of non-model species is improved, thereby establishing a marker gene system belonging to the cells of non-model species themselves. On the other hand, the ligand-receptor database of model species is applied to the cell communication analysis of non-model species, making the cell communication analysis of non-model species possible, and realizing a breakthrough from 0 to 1.

[0051] Example 2

[0052] In this embodiment, according to a non-model species cell annotation method for bidirectional homology alignment provided in Example 1, after bidirectional homology alignment of the rat genome to the mouse genome, the cell annotation result of the rat is obtained, specifically, Figure 2 This is a t-SNE diagram of manually annotated cells in rats; Figure 3 This is the t-SNE diagram of the bidirectional homology alignment of the rat genome blast to the mouse. Since the rat gene has an ensembl ID number, Figure 4 It is a t-SNE diagram of the bidirectional homology alignment of the rat genome ensembl to the mouse. The annotation results of the rat cell types are counted according to the cell types, and the distribution of the rat cell types is visualized through a table. The results are described in Table 1. It should be noted that manual annotation is to determine the cell type through subjective observation of the distribution of marker genes, thereby annotating the cells of non-model species, which is the baseline for the credibility of cell annotation results at this stage.

[0053] Table 1 Statistics of cell annotation results of rat genome bidirectional homology alignment to mouse

[0054]

[0055] Combination Figure 2 , Figure 3 and Figure 4From the statistics of the cell annotation results of the rat genome bidirectional homology alignment to the mouse in Table 1, it can be seen that, since rats and mice are closely related, the results of the cell type annotation obtained by manual annotation and by bidirectional homology alignment are almost the same, with no difference. The cell type annotation result obtained by the bidirectional homology alignment only has one more cell type than the manual annotation result; the number of cells aligned by the ensembl homology alignment method is more than the number of cells aligned by the blast homology alignment method, but the cell type annotation results finally obtained by the homology alignment method are almost the same, with no difference, indicating that the statistical results of the annotation of non-model species cells by bidirectional homology alignment in the present invention are also credible.

[0056] Example 3

[0057] In this embodiment, according to a non-model species cell annotation method for bidirectional homology alignment provided in Example 1, after bidirectional homology alignment of the rat genome to the human genome, the single cell annotation result of the rat is obtained, specifically, Figure 2 This is a t-SNE diagram of manually annotated cells in rats; Figure 5 This is the t-SNE diagram of the rat genome blast bidirectional homology alignment to the human genome; since the rat gene has an ensembl ID number, Figure 6 This is the t-SNE diagram of the rat genome ensembl bidirectional homology alignment to the human genome. The annotation results of the rat cell types are counted according to the cell types, and the distribution of the rat cell types is visualized through a table. The results are shown in Table 2.

[0058] Table 2 Statistics of cell annotation results of bidirectional homology alignment of rat genome to human rat genome

[0059]

[0060]

[0061] Combination Figure 2 , Figure 5 and Figure 6 From the statistics of the cell annotation results of the rat genome bidirectional homology alignment to that of humans in Table 2, it can be seen that due to the distant relationship between rats and humans, the results of the cell type annotation obtained by manual annotation and by bidirectional homology alignment are very different. The cell type annotation results obtained by bidirectional homology alignment include 23 types of cells, while the manual annotation results only include 9 types of cells. The number of cells aligned by the ensembl homology alignment method is more than that by the blast homology alignment method, but the cell type annotation results finally obtained by the homology alignment method are almost the same, with no difference.

[0062] Example 4

[0063] Download the original sequencing data of rat single-cell transcriptome with GSE number GSE165153 published in the high-throughput gene expression database (Gene Expression Omnibus, GEO) in the article "Single-cell profiling of CNS border compartment leukocytes reveals that B cells and their progenitors reside in non-diseased meninges" from the journal Nature Neuroscience, and use the non-model species cell annotation method of bidirectional homology alignment provided by the present invention to perform cell type annotation. The obtained UMAP map is shown in the attached figure. Figure 7 The original UMAP map of rats versus humans obtained in the article is shown in the attached Figure 8 As shown, although the UMAP map obtained in this example is different from that in the article, from the cell type annotation results, according to the description of the article, Figure 7 Myeloid1 in Figure 8 Macro, Figure 7 Myeloid2 in Figure 8 CAM, Figure 7 Myeloid3 in Figure 8 Mono, the cell type annotation results obtained by the non-model species cell annotation method for bidirectional homology comparison provided by the present invention are consistent with those in the article, indicating that the annotation results of the non-model species cell annotation method for bidirectional homology comparison provided by the present invention are accurate. It should be noted that although the UMAP graph obtained in this embodiment is different from that in the article, the reason is that in the dimensionality reduction process, when reducing high-dimensional data to low-dimensional space, UMAP will select several random points in the low-dimensional space as starting points, and then calculate the low-dimensional coordinates corresponding to other high-dimensional space points based on the coordinates of these points. It is precisely the existence of this randomness that for the same data, when performing two dimensionality reductions, the calculated coordinate values ​​will be different. Although this embodiment will set a random number seed on the same computer to ensure that the two dimensionality reductions use a fixed starting random number, thereby ensuring that the two dimensionality reduction graphs are the same, the data of the article is analyzed using different computers than this embodiment, and the random number will change, resulting in the UMAP graph obtained in this embodiment being different from that in the article, but the cell type annotation results obtained in this embodiment are consistent with those in the article.

[0064] Example 5

[0065] Fig. 9 This is the UMAP map of rats and humans obtained by the non-model species cell annotation method of bidirectional homology alignment used in this embodiment, combined with the attached Figure 8 Regarding the UMAP diagram of rats compared with humans, it can be seen from the figure that although the UMAP diagram obtained in this embodiment and the UMAP diagram obtained in the article are different, in fact, the difference between the two diagrams is not large. It is just that the article and this embodiment use different marker gene databases. The article uses the bone marrow single cell transcriptome sequencing data of Tabula Muris with GSM number GSM2967055. Figure 8 Bc1 and Bc2 are merged into Fig. 9 B cells, Figure 8 mDC1 and pDC were combined into Fig. 9 Dendritic cells, Figure 8 Myeloid1 / 2 / 3, granulo and micro are combined into Fig. 9 The monocytes, among which Figure 8 Myeloid 1 / 2 / 3, granulo and micro are all myeloid cells, but the myeloid cell in the human marker gene database of this embodiment only has monocytes. Fig. 9 CD4 Tc is renamed as CD4+T cells. Fig. 9 The CD8 Tc was renamed as CD8+ T cells. Fig. 9 The NK cells are split into NK cells and NK T cells, so the UMAP graph obtained in this example is different from the UMAP graph obtained in the article. Figure 1 To.

[0066] Example 6

[0067] Fig.10 is a UMAP map of rats versus mice obtained by the non-model species cell annotation method using bidirectional homology alignment in this embodiment, Fig.11 It is a Sankey diagram, which reflects the relationship between the cell type annotation results obtained by the annotation method of this embodiment and the article. It can be seen from the figure that the cell type annotation results obtained by aligning the rat genome to the human and mouse databases are affected by the type of marker gene database, but there is actually no difference in accuracy. It is precisely because of the differences between the databases that when actually annotating cell types for non-model species, it is necessary to call the marker gene database that conforms to the non-model species.

[0068] In summary, the credibility of comparing and annotating non-model species in Examples 2 and 3 is still relatively high, but it still depends on the degree of closeness with the model species; the number of cells compared by the homology comparison method of ensembl is more than the number of cells compared by the homology comparison method of blast, but the difference in the results of cell type annotation obtained by the homology comparison method is very small. The specific application depends on whether the research species has an ensembl ID number. If so, it is recommended to use the ensembl comparison method, which has high comparison coverage and fast calculation speed; if not, the blast comparison method is used, which has a wide range of applications and high flexibility. In Examples 4 and 5, the cell type annotation results obtained by bidirectional homology comparison of the rat genome with the genomes of humans and mice are relatively accurate, and are consistent with the cell type annotation results obtained in the article, indicating that the non-model species cell annotation method for bidirectional homology comparison provided by the present invention can be applied to the marker gene database of humans and mice, thereby improving the scope of application in practical applications.

[0069] The above-mentioned embodiments are only preferred embodiments of the present invention and cannot be used to limit the scope of protection of the present invention. Any non-substantial changes and substitutions made by technicians in this field on the basis of the present invention shall fall within the scope of protection required by the present invention.

Claims

1. A non-model species cell annotation method for bidirectional homology alignment, characterized in that: The following steps are involved: S1. Perform single-cell transcriptome sequencing on non-model species to obtain single-cell transcriptome sequencing data; S2, aligning the genome of the non-model species with the genome of the model species, and retaining the only aligned gene pairs; S3, aligning the genome of the model species with the genome of the non-model species, and retaining the only aligned gene pairs; S4, comparing the uniquely aligned gene pairs retained in S2 with the uniquely aligned gene pairs retained in S3, retaining the same gene pairs as the gene pairs for bidirectional homologous alignment; S5. In the marker gene database of the model species, the genes of the non-model species in the gene pairs of the bidirectional homologous alignment are replaced by the genes of the non-model species in the gene pairs of the bidirectional homologous alignment to obtain the expression matrix of the non-model species; In S5, the marker gene database of the model species includes an expression matrix of the model species, in which the gene ID of the model species in the gene pair of bidirectional homologous alignment is replaced with a gene homologous to the model species in the gene pair of bidirectional homologous alignment, to obtain the expression matrix of the non-model species; S6. The gene data of the expression matrix of the non-model species is called by software for annotation to obtain annotation results of the cell type of the non-model species.

2. A non-model species cell annotation method for bidirectional homology alignment according to claim 1, characterized in that: In S2 and S3, the genome of the model species and the genome of the non-model species are compared using the blast method.

3. A non-model species cell annotation method for bidirectional homology alignment according to any one of claims 1 to 2, characterized in that: The singleR software is used to call the gene data of the expression matrix of the non-model species for annotation, and the annotation results of the cell type of the non-model species are obtained.

4. The method for annotating non-model species cells by bidirectional homology alignment according to claim 1, characterized in that: After obtaining the annotation results of the non-model species cell type, using the ligand-receptor database of the model species, according to the obtained annotation results of the non-model species cell type, cell communication analysis can be performed on the non-model species cells based on the annotated non-model species cell type.

5. The method for annotating non-model species cells by bidirectional homology alignment according to claim 1, characterized in that: After obtaining the annotation results of the non-model species cell types, the annotation results of the non-model species cell types are counted according to the cell types, and the distribution of the non-model species cell types is visualized through a table.

6. A non-model species cell annotation method for bidirectional homology alignment according to any one of claims 1 to 2, characterized in that: The marker gene database of the model species includes a marker gene database of humans and a marker gene database of mice.

Images