WO2021253544A1 - Model using 87 genes serving as biomarkers to predict cell proliferation activity - Google Patents

Abstract

Provided is a model using 87 genes serving as biomarkers to predict cell proliferation activity. The expression level of a set of cell proliferation genes correlates positively with cell proliferation activity. The present invention provides a set of methods to evaluate cell proliferation activity without requiring in vitro culture. Combined with single-cell sequencing, the model can measure, quickly and easily, the proliferation activity of various in vivo cell types, determining whether cancer tissues have normal cells that are significantly proliferating.

Description

A model for predicting cell proliferation activity using 87 genes as biomarkers Technical field

The invention belongs to the fields of gene technology and biomedicine, and specifically relates to a method for predicting cell proliferation activity using 87 genes as biomarkers

Background technique

The disorderly proliferation of cancer cells is a key mechanism for tumorigenesis. In view of the cell proliferation mechanism, treatment methods such as chemotherapy have been developed. At the same time, a number of cell proliferation gene markers such as MKI67, MCM2 and PCNA have been developed to use their mRNA or protein expression levels to indicate the proliferation activity of cancer cells, thereby assisting in evaluating the prognosis of patients after surgery. Especially for the protein expression level of MKI67, the Ki-67 index was developed to mark the ratio of Ki-67 positive cells in pathological samples, so as to evaluate lung cancer, breast cancer, prostate cancer, cervical cancer, colorectal cancer, bladder cancer, The prognosis of cancer patients such as lymphoma.

Proliferation is not unique to cancer cells. Studies have shown that there are a large number of cells with proliferation activity in human skin, bone marrow and gastrointestinal tissues. When cancer occurs in the above-mentioned tissues, the expression of cell proliferation markers such as MKI67 in cancer tissue samples of patients after surgery is partly derived from cancer cells and partly derived from normal proliferating cells, which will not accurately reflect the proliferation activity of cancer cells. Due to the lack of sufficient data support, the American Society of Clinical Oncology (ASCO) Tumor Marker Steering Committee does not recommend the Ki-67 index as a routine prognostic marker for patients newly diagnosed with breast cancer. Part of the reason for this phenomenon is that there are also a large number of proliferating cells in immune organs such as normal bone marrow and lymph nodes. In the pathological samples of patients, the Ki-67 index cannot accurately distinguish between normal proliferating cells and tumor cells, resulting in the accuracy of estimating the proliferation activity of cancer cells. Decrease, which in turn leads to a decline in the ability to predict the prognosis of patients.

In vitro culture can help us identify the proliferation ability of normal cells. However, this method currently has great difficulties: 1. Some cells cannot be cultured in vitro; 2. Due to huge differences in the living environment of some cells, their proliferation ability under in vitro culture conditions cannot reflect the true proliferation ability in the in vivo environment.

Summary of the invention

In view of the differences in the proliferation activity of different types of cells in the human body and the difficulty in evaluating cell proliferation activity in current culture methods, the present invention provides a method for evaluating cell proliferation activity using a collection of 87 cell proliferation genes as markers. In order to achieve this objective, the present invention adopts the following technical solutions.

1. Establish the cell proliferation gene collection, which consists of 87 genes. The specific implementation steps are as follows:

(1) Data collection

Obtain single-cell RNA-Seq data of different types of normal cells from the Tabula Muris database (https://tabula-muris.ds.czbiohub.org/), from the Cancer Genome Atlas (TCGA) database (http://cancergenome.nih. gov/) Obtain the RNA-Seq data of cancer and adjacent tissues, obtain the tissue RNA-Seq data from the GTEx (Genotype-Tissue Expression Project) database (https://www.gtexportal.org/), and obtain the tissue RNA-Seq data from CCLE (Cancer Cell Line) Encyclopedia) database (https://portals.broadinstitute.org/ccle) to obtain cell line RNA-Seq data and cell proliferation activity data.

(2) Mining the collection of specific expression genes in stem/group cells

a) The normal single cells in the body in the Tabula Muris database are classified into 81 types according to cell types, and the gene expression values of various types of cells are calculated. For a certain gene j in a certain cell type i, calculate its expression value (X _ji ) as follows:

Where m is the total number of cells belonging to cell type i, and n is the number of cells whose read count of cell gene j in cell type i is greater than 0. In this way, the expression values of all genes in cell type i are calculated. In turn, the expression values of all genes of 81 cell types are calculated.

b) Divide 81 types of cells into two groups: stem/group cell group and other cell group.

c) Use hierarchical clustering analysis to mine the genes that are highly expressed in the stem/group cell group and extremely lowly expressed in other cell groups as a set of stem/group cell-specific genes.

(3) Mining of cell proliferation gene collection

a) Obtain the expression values of genes in the stem/group cell specific gene set in each normal tissue sample in the GTEx database. In most normal tissues, terminal cells that do not have proliferative activity occupies the main component. For this reason, the above-mentioned genes were analyzed by hierarchical clustering, and a gene group consisting of 87 genes that were low expressed in normal tissues was obtained (87 genes including ANLN, ARHGAP11A, ASF1B, ATAD2, AURKA, AURKB, BIRC5, BRCA2, BUB1, BUB1B, CCNA2, CCNB1, CCNB2, CDC20, CDC45, CDCA2, CDCA5, CDCA8, CDK1, CDT1, CENPA, CENPE, CENPF, CENPH, CENPK, CENPM, CENPW, CEP55, CKAP2, CKAP2L, CLSPN, DBF4, DLGAP5, ECT2, ESCO2, FEN1, FOXM1, HIRIP3, HIST1H2AG, HMMR, KIF11, KIF15, KIF20A, KIF20B, KIF23, KIFC1, LMNB1, LMNB2, LLR1, MAD2L MCM2, MKI67, NCAPG, NCAPG2, NCAPH, NDC80, NEIL3, NUDT1, NUF2, NUSAP1, PBK, PKMYT1, PLK1, PLK4, PRC1, RACGAP1, RAD51, RCC1, RRM2, SHCBP1, SKA1, SMC2, SNRNP25, SPC24, SPC25, SYCE2, TACC3, TK1, TOP2A, TPX2, TRIM59, TRIP13, TYMS, UBE2C, UBE2T, UHRF1 and VRK1).

b) Obtain the expression values of the above 87 genes in cancer and adjacent tissue samples from the TCGA database. For a certain gene j (1≤j≤87), calculate its Z-score normalized gene expression value Y _j in all cancer and para-cancerous tissue samples. For a certain sample k, enumerate the expression vectors of its 87 genes as {Y _1k ,Y _2k ,...,Y _87k }, and then calculate the expression value of the gene set as the median value of the above 87 gene expression vectors (median{Y _1k ,Y _2k ,…,Y _87k }). The T test is further used to compare the gene set expression values of each cancer sample with the gene set expression values of all adjacent samples. Since most cancer tissues are composed of highly proliferative cancer cells, it is further confirmed that the above-mentioned gene set is highly expressed in cancer tissues and lowly expressed next to cancer. So far, it has been confirmed that the above-mentioned 87 genes are a cell proliferation gene set.

2. Cell proliferation genes using the established set of prediction model of cell proliferation activity, specific implementation steps are as follows:

(1) The cell proliferation gene set predicts the proliferation activity of cancer cell lines in vitro

a) Obtain the expression values of genes in the cell proliferation gene set in each cancer cell line in the CCLE database. Similarly, for a certain gene j (1≤j≤87), calculate its Z-score normalized gene expression value Z _j in all cell line samples. For a certain cell line sample k, enumerate the expression vectors of 87 genes as {Z _1k ,Z _2k ,...,Z _87k }, and then calculate the expression value of the gene set as the median value of the 87 gene expression values (median {Z _1k ,Z _2k ,…,Z _87k }). Calculate the expression value of the cell proliferation gene set for each cell line sample.

b) Obtain some cell proliferation activity data (doubling time) in the CCLE database.

c) Perform Pearson correlation analysis on the expression value data of the cell proliferation gene set of the cell line sample and the doubling time data of the corresponding cell line. It is confirmed that in cancer cell lines derived from solid tumors, cell proliferation activity is significantly positively correlated with the expression value of the cell proliferation gene set composed of 87 genes, that is, the expression level of the cell proliferation gene set can predict the cancer cell line derived from solid tumors. Proliferative activity.

(2) Establish a prediction model for cell proliferation activity

a) The single cells in the Tabula Muris database are classified into 81 categories by cell type, and the gene expression values of various types of cells are obtained as above.

b) Perform hierarchical cluster analysis on 81 cell types using the expression values of the above 87 genes. Through cluster analysis, the cell types are grouped into 2-3 categories.

c) For each of the 81 cell types, calculate the expression value of the cell proliferation gene set, and obtain the expression value of 87 genes in the cell proliferation gene set in each cell type. For a certain cell type i, for a certain gene j (1≤j≤87 gene expression value X _ji , enumerate the expression vector of 87 genes as {X _1i ,X _2i ,...,X _87i }, and then calculate The expression value of the cell proliferation gene set is the median value of the above 87 gene expression values (median{X _1i , X _2i ,..., X _87i ).

d) According to the results of cluster analysis, 81 cell types are grouped into 2-3 different cell type groups, for each cell type group, the expression value vector of its cell proliferation gene set is obtained, and the different cell type groups are compared. The expression value of the cell proliferation gene set (T test, two-tailed). Using P<0.05 as the threshold, judge whether the expression value of the cell proliferation gene set of a certain cell type group is significantly higher than that of other cell type groups, so as to confirm the cell type with high expression of cell proliferation gene and the cell type with low expression of cell proliferation gene. Evaluation of the proliferation activity of 81 cell types.

So far, the expression level of the cell proliferation gene set has been used to evaluate the proliferation activity of 81 normal cell types in vivo.

The present invention uses single-cell RNA-Seq data to identify a cell proliferation-related gene marker set composed of 87 genes. Using this set, we evaluate the proliferation activity of different normal cell types in the body, and comprehensively identify the normal cell types that proliferate at high speed in the body. The realization of this technology can help us determine whether there are normal cells with proliferation activity in cancer tissues. When there are a large number of such cells in cancer tissues, the treatment and evaluation methods for cell proliferation markers will be interfered and may fail.

The advantages of the present invention are that (1) The method for judging cell proliferation activity based on culture requires in vitro culture of normal tissue cells. At present, some tissue cells cannot be cultured in vitro, and some tissue cells are affected by culture conditions. There are huge differences in cell proliferation activity in vivo and in vitro. This method uses single-cell technology to directly evaluate the proliferation activity of tissue cells in the body, and can accurately evaluate the proliferation activity of cells. (2) The results of normal cell proliferation activity obtained by this method can assist in determining whether there are a large number of normal cells with proliferation ability in cancer tissues, so as to provide guidance for cancer treatment and evaluation methods for cell proliferation mechanisms.

Description of the drawings

Figure 1: Heat map of cluster analysis of high-expressed genes in stem/group cell group. In the figure, one column represents a cell type, and one row represents a gene. Perform cluster analysis on genes whose expression level of any cell type in the stem/group cell group is greater than 0.5, cluster them into 15 gene groups, and find a gene group consisting of 162 genes, whose genes are significantly expressed in the stem/group cell group , Very low expression in other cell types. In the figure, Epi-SC indicates epidermal stem cells, numbers 1-7 indicate Slamf1 positive pluripotent cells (1), megakaryocyte-erythroid progenitor cells (2), advanced B precursor cells (3), and granulomanocytic cells (4) ), granulocytes (5), lymphoid progenitor cells (6) and natural killer precursor cells (7), these 8 types of cells form a stem/group cell group.

Figure 2: Heat map of cluster analysis of stem/group cell-specific gene set genes in 54 normal tissue samples. In the figure, one column represents one sample, one row represents one gene, and samples of the same color belong to the same tissue type. Cluster analysis was performed on 17382 samples of 54 normal tissues of stem/group cells specifically expressed gene set, and clustered into 2 gene clusters. It was found that the gene cluster formed by the aggregation of 87 genes was only in (1) cultured skin fibroblasts after culture, (2) EBV-transformed lymphocytes (EBV-transformed lymphocytes) and (3) testis tissue (testis) High expression in tissues, low expression in other tissues.

Figure 3: Box plot of expression levels of cell proliferation gene sets in different cancers. Obtain the expression level of the cell proliferation gene set of 9630 samples from 32 kinds of cancer and adjacent tissues, and then merge all the adjacent samples (Control). The t-test was used to compare the expression levels of cell proliferation gene sets for each cancer and Control. Taking the two-tailed P-value<0.05 as an indicator, the expression level of the cell proliferation gene set in the red highlight was significantly higher than that of the cancer types in the adjacent group.

Figure 4: Correlation analysis between the expression level of cell proliferation gene set and the optimal doubling time of cells. Each point in the figure represents a cell line, the abscissa represents the expression level of the cell line's cell proliferation gene set, and the ordinate represents the doubling time of the cell line (provided by the supplier). Calculate the Pearson correlation coefficient and P-value of the cell proliferation gene set expression level and the optimal cell doubling time.

Figure 5: Cluster analysis heat map of 81 different normal cell types in Tabula Muris. In the figure, one column represents a cell type, and one row represents a gene in a cell proliferation gene set. According to the expression levels of genes in the cell proliferation gene set, 81 normal cell types are grouped into three types.

detailed description

The present invention will be described in detail below with reference to the accompanying drawings and embodiments. The following are only preferred embodiments of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the method of the present invention, they can also make Several improvements and supplements, these improvements and supplements should also be regarded as the protection scope of the present invention.

Example 1: Using Tabula Muris database, TCGA database, GTEx database and CCLE database to establish a cell proliferation gene set containing 87 genes, predict the proliferation activity of 81 different normal cell types in the body collected in the Tabula Muris database, and assist in determining TCGA Whether there are a large number of normal cells with proliferation activity in the cancer tissues in the database, guide the cancer treatment and evaluation methods for cell proliferation markers.

(1) Data collection

Obtained 53760 single-cell RNA-Seq data from 81 different normal cell types generated by Smart-Seq2 single-cell sequencing technology from Tabula Muris database. Obtained from the Cancer Genome Atlas (TCGA) database, 9630 cancer and adjacent tissue RNA-Seq data of 32 cancers, and prognostic data of 31 cancers. 17382 tissue RNA-Seq data of 54 tissues were obtained from GTEx database. The RNA-Seq data of 1019 cell line samples and the culture mode (suspension/adherent/semi-adherent) and doubling time information of some of the cell lines were obtained from the CCLE database.

(2) Mining the collection of specific expression genes in stem/group cells

First, we classified the normal single cells in the 53760 individuals collected in the Tabula Muris database into 81 types according to cell types, and calculated the expression values of genes in different cell types. For a certain gene j in a certain cell type i, calculate its expression value (X _ji ) as follows:

Where m is the total number of cells belonging to cell type i, and n is the number of cells in cell type i whose read count of cell gene j is greater than 0. In this way, the expression values of all genes in cell type i are calculated. In turn, the expression values of all genes of 81 cell types are calculated.

Secondly, the 81 types of cells are divided into two groups: stem/group cell group and other cell group (Table 1). The stem/group cell group includes epidermal stem cells (stem cells of epidermis), Slamf1-positive multipotent progenitor cells, megakaryocyte-erythroid progenitor cells, and advanced B precursor cells ( late pro-B cell, granulocyte monocyte progenitor cell, granulocytopoietic cell, common lymphoid progenitor and pre-natural killer cell; others The cell group includes the remaining 73 types of cells.

Finally, select genes whose expression value is greater than 0.5 in any cell type of the stem/group cell group. For these genes, hierarchical cluster analysis was used to mine a gene group composed of 162 genes that are highly expressed in the stem/group cell group while being extremely low in other cell groups, as a stem/group cell-specific gene set (Figure 1).

Table 1: 81 normal cell types in Tabula Muris database

(3) Mining of cell proliferation gene collection

First, obtain the expression values of 162 genes in the stem/group cell-specific gene set in each normal tissue sample in the GTEx database. For each gene, its expression value in 17,382 tissue samples was obtained, and Z-score normalization was performed to obtain the normalized expression value of the gene. Accordingly, the normalized expression values of 162 genes were obtained. Hierarchical clustering analysis of genes, since terminal cells that do not have proliferative activity in most normal tissues occupy the main component, only (i) cultured skin fibroblasts (cultured skin fibroblasts), (ii) EBV are obtained after culture Gene clusters formed by 87 genes that are highly expressed in transfected lymphocytes (EBV-transformed lymphocytes) and (iii) testis tissues (testis), and lowly expressed in 51 other tissues (Figure 2). These 87 genes constitute the cell proliferation gene set (Table 2).

Table 2: Gene list of cell proliferation gene set

ANLN	CCNA2	CENPA	CLSPN	KIF11	MCM2	PBK	SKA1	TRIM59
ARHGAP11A	CCNB1	CENPE	DBF4	KIF15	MKI67	PKMYT1	SMC2	TRIP13
ASF1B	CCNB2	CENPF	DLGAP5	KIF20A	NCAPG	PLK1	SNRNP25	TYMS
ATAD2	CDC20	CENPH	ECT2	KIF20B	NCAPG2	PLK4	SPC24	UBE2C
AURKA	CDC45	CENPK	ESCO2	KIF23	NCAPH	PRC1	SPC25	UBE2T
AURKB	CDCA2	CENPM	FEN1	KIFC1	NDC80	RACGAP1	SYCE2	UHRF1
BIRC5	CDCA5	CENPW	FOXM1	LMNB1	NEIL3	RAD51	TACC3	VRK1
BRCA2	CDCA8	CEP55	HIRIP3	LMNB2	NUDT1	RCC1	TK1	To
BUB1	CDK1	CKAP2	HIST1H2AG	LRWD1	NUF2	RRM2	TOP2A	To
BUB1B	CDT1	CKAP2L	HMMR	MAD2L1	NUSAP1	SHCBP1	TPX2	To

Secondly, the expression values of the above 87 genes in a total of 9630 samples from 32 cancers (Table 3) in cancer tissues and adjacent tissues in the TCGA database were obtained. For a certain gene j (1≤j≤87), calculate its Z-score normalized gene expression value Y _j in all cancer and para-cancerous tissue samples. For a certain sample k, enumerate the expression vectors of its 87 genes as {Y _1k ,Y _2k ,...,Y _87k }, and then calculate the expression value of the gene set as the median value of the above 87 gene expression vectors (median{Y _1k ,Y _2k ,…,Y _87k }). The T test is further used to compare the gene set expression values of each cancer sample with the gene set expression values of all adjacent samples. Since most cancer tissues are composed of highly proliferating cancer cells, it is further confirmed that in 28 cancers of 32 cancer types, the expression value of cell proliferation gene sets in cancer tissues is higher than that in adjacent tissues (P<0.05) , Table 3 and Figure 3).

Table 3: Chinese and English comparison of the names of 32 cancers in the TCGA database

HNSC	Head and neck squamous cell carcinoma	SKCM	Skin melanoma
KICH	Chromophobe renal cell carcinoma	STAD	Stomach cancer
KIRC	Renal clear cell carcinoma	BLCA	Bladder urothelial carcinoma
KIRP	Papillary cell carcinoma of the kidney	TGCT	Testicular cancer
LAML	Acute myeloid leukemia	THCA	Thyroid cancer
LGG	Low-grade glioma of the brain	THYM	Thymic Cancer
LIHC	Hepatocellular carcinoma	UCEC	Endometrial cancer
LUAD	Lung adenocarcinoma	UCS	Uterine Sarcoma
LUSC	Lung squamous cell carcinoma	UVM	Uveal melanoma
ACC	Adrenocortical carcinoma	BRCA	Invasive breast carcinoma
MESO	Mesothelioma	CESC	Cervical squamous cell carcinoma and adenocarcinoma
OV	Serous cystadenocarcinoma of ovary	COAD	Colon cancer
PAAD	Pancreatic cancer	DLBC	Diffuse large B cell lymphoma
PCPG	Pheochromocytoma and paraganglioma	ESCA	Esophageal cancer
PRAD	Prostate cancer	GBM	Multiforme glioma
READ	Rectal adenocarcinoma	SARC	sarcoma

(4) Cell proliferation gene set predicts cancer cell proliferation activity

First, obtain the expression values of 87 genes in the cell proliferation gene set of all cancer cell lines in the CCLE database.

For a certain gene j (1≤j≤87), calculate its Z-score normalized gene expression value Z _j in 1019 cell line samples. For a certain cell line sample k, enumerate the expression vectors of 87 genes as {Z _1k ,Z _2k ,...,Z _87k }, and then calculate the expression value of the gene set as the median value of the 87 gene expression values (median {Z _1k ,Z _2k ,…,Z _87k }).

Calculate the expression value of the cell proliferation gene set for each cell line sample.

Secondly, the CCLE database provides information on the culture method (suspension/semi-adherent/adherent) and doubling time (provided by the supplier/statistic by the CCLE staff) of some cell lines. It is believed that the doubling time provided by the supplier expresses the optimal doubling time of the cell line and can be used as an indicator of the cell line's cell proliferation activity. To this end, data on the doubling time of 99 semi-adherent or adherent cell lines (provided by the supplier) were obtained as an indicator of the cell proliferation activity of this batch of cell lines. Finally, it was found that the expression value of the cell proliferation gene set of this batch of cell lines was negatively correlated with the cell doubling time (Pierce correlation analysis, Figure 4). As the cell doubling time is shorter, the cell proliferation activity is stronger. This result indicates that there is a positive correlation between the expression value of the cell proliferation gene set and the cell proliferation activity. Generally, solid tumors grow in a semi-adherent or adherent manner, while hematomas grow in a suspended manner. This result indicates that the expression value of the cell proliferation gene set of solid tumors can predict its cell proliferation activity.

(5) Evaluation of cell proliferation activity

First, the single cells in the Tabula Muris database are classified into 81 categories by cell type, and the gene expression values of various types of cells are calculated as above. For these 81 different cell types, the expression values of 87 genes in the cell proliferation gene set in each cell type were obtained.

Secondly, using the expression values of the 87 genes mentioned above to perform hierarchical cluster analysis on 81 different normal cell types. Use the R software package "factoextra" to perform hierarchical clustering analysis. The distance metric used is "euclidean" and the clustering method is "ward.D2". According to the results of the hierarchical clustering tree, 81 cell types are grouped into three categories (Figure 5). One type is the stem/group cell group, and the other two types are derived from other cell groups (Table 1), namely the significant proliferation group (the cells in this group have significant cell proliferation gene expression and thus have certain cell proliferation ability) and the rare proliferation group ( (This group of cells rarely express cell proliferation genes and therefore have weak cell proliferation ability).

Finally, for each of the 81 cell types, the expression value of its cell proliferation gene set was calculated. For each of the 81 cell types, the expression value of its cell proliferation gene set was calculated. Obtain the expression values of 87 genes in the cell proliferation gene set in each cell type. For a certain cell type i, for a certain gene j (1≤j≤87 gene expression value X _ji , enumerate the expression vector of 87 genes as {X _1i ,X _2i ,...,X _87i }, and then calculate The expression value of the cell proliferation gene set is the median value of the above 87 gene expression values (median{X _1i , X _2i ,..., X _87i ). Compare the cell proliferation gene set expression values of the above three types of different cell type groups. Use. The T test method, using the two-tailed P<0.05 as the threshold, found that the expression value of the cell proliferation gene set of the stem/group cell group was significantly greater than the expression value of the cell proliferation gene set of the significant proliferation group, and the expression of the cell proliferation gene set of the significant proliferation group The value is greater than the expression value of the cell proliferation gene set of the rare proliferation group. In this way, 81 different normal cell types are successfully divided into three types of cell type groups with different cell proliferation capabilities, and the level evaluation of the proliferation ability of the corresponding cell types is achieved.

(6) Clinical application of normal cell type proliferation activity to guide cell proliferation markers

First of all, based on cell type information, it was found that in the significant proliferation group, immature B cell, basal cell of epidermis (epidermis basal cell), epidermal cell of large intestine (large intestine epithelial cell) may have solid tumors in the tissues. .

Secondly, the analysis of 31 groups of solid tumors with clinical prognostic information in TCGA found that DLBC (diffuse large B cell lymphoma) cancer tissue may contain a large number of immature B cells, HNSC (head and neck squamous cell carcinoma), LUSC (lung Squamous cell carcinoma), ESCA (esophageal cancer) and CESC (cervical squamous cell carcinoma and adenocarcinoma) all contain squamous cell carcinoma, the cancerous tissue may contain a large number of epidermal basal cells, while COAD (colon cancer) and READ (rectal adenocarcinoma) The cancer tissue may contain a large number of large intestinal epithelial cells. These 7 types of cancers all contain a large number of normal cells with significant proliferation activity, and treatment and prognosis prediction based on cell proliferation markers may fail.

Finally, using TCGA's cancer tissue RNA-Seq data and clinical progression-free interval (PFI, disease remission) data, using Cox proportional hazard regression model (continuous variable method) method to determine DLBC, HNSC, LUSC, ESCA, CESC, Whether the expression value of proliferation marker MKI67 in cancer tissue samples after COAD and READ patients has predictive significance for their disease remission. Using P-value<0.05 as the threshold, it was found that the expression value of MKI67 could not predict the remission period of these 7 cancers (Table 4). This result is consistent with our forecast.

Table 4: Prognostic analysis of cancer cell proliferation marker MKI67, a cancer cell proliferation marker with a large number of significantly proliferating normal cell types in 7 cancer tissues

Cancer type	Hazard Ratio (95% confidence interval)	Type 3 P-value
Cervical squamous cell carcinoma and adenocarcinoma	1.03(0.77-1.37)	0.8566
Colon cancer	0.87(0.54-1.4)	0.5555
Diffuse large B cell lymphoma	1(0.55-1.81)	0.9902
Esophageal cancer	1.04(0.89-1.21)	0.6189
Head and neck squamous cell carcinoma	0.97 (0.84-1.12)	0.6502
Lung squamous cell carcinoma	1.18 (0.95-1.45)	0.1277
Rectal adenocarcinoma	1.61(0.73-3.55)	0.242

Claims (4)

1. A model for predicting cell proliferation activity using 87 genes as biomarkers is characterized in that it is achieved through the following steps:

   (1) Establish a collection of cell proliferation genes:

   1) Data collection

   Obtain single-cell RNA-Seq data of different types of normal cells from the Tabula Muris database, RNA-Seq data of cancer and para-cancerous tissues from the Cancer Genome Atlas database, tissue RNA-Seq data from the GTEx database, and cell lines from the CCLE database RNA-Seq data and cell proliferation activity data;

   2) Mining the collection of specific expression genes in stem/group cells

   a) The normal single cells in the body in the Tabula Muris database are classified into 81 types by cell type, and the gene expression value of each type of cell is calculated. For a certain gene j in a specific cell type i, the expression value (X _ji )as follows:

   

   Where m is the total number of cells belonging to cell type i, n is the number of cells whose read count of cell gene j in cell type i is greater than 0, calculate the expression values of all genes in cell type i, and count all genes of 81 cell types in turn Expression value of

   b) Divide 81 types of cells into two groups: stem/group cell group and other cell groups;

   c) Use hierarchical clustering analysis to mine the genes that are highly expressed in the stem/group cell group and extremely lowly expressed in other cell groups as a set of stem/group cell-specific genes;

   3) Mining of cell proliferation gene collection

   a) Obtain the expression values of genes in the stem/group cell-specific gene set in each normal tissue sample in the GTEx database. In most normal tissues, terminal cells that do not have proliferative activity occupy the main components, and the genes are hierarchized for this purpose Cluster analysis to obtain a gene group consisting of 87 genes that are low expressed in normal tissues;

   b) Obtain the expression values of the above 87 genes in the cancer and adjacent tissue samples from the TCGA database. For a certain gene j, calculate the Z-score normalized gene expression value Y _{j in all cancer and adjacent tissue samples.} For a sample k, enumerate the expression vectors of its 87 genes as {Y _1k ,Y _2k ,...,Y _87k }, then calculate the expression value of the gene set as the median value of the 87 gene expression vectors, and further use the T test to The gene set expression value of a cancer sample is compared with the gene set expression values of all adjacent samples. Since most cancer tissues are composed of highly proliferating cancer cells, it is further confirmed that the above gene sets are highly expressed in cancer tissues. Side low expression, confirming that the gene group composed of the above 87 genes is a cell proliferation gene collection;

   (2) Use the above cell proliferation gene set to establish a model for predicting cell proliferation activity:

   1) Cell proliferation gene set predicts the proliferation activity of cancer cell lines in vitro

   a) Obtain the expression values of genes in the cell proliferation gene set in each cancer cell line in the CCLE database. For a certain gene j, calculate the Z-score normalized gene expression value Z _j in all cell line samples. For a certain cell Line sample k, enumerate the expression vector of its 87 genes as {Z _1k ,Z _2k ,...,Z _87k }, and then calculate the expression value of the gene set as the median value of the above 87 gene expression values, and calculate each cell line The expression value of the cell proliferation gene set of the sample;

   b) Obtain some cell proliferation activity data in the CCLE database;

   c) Perform Pearson correlation analysis on the expression value data of the cell proliferation gene set expression value of the cell line samples and the doubling time data of the corresponding cell line, and confirm that the cell proliferation activity is composed of 87 genes in the cancer cell line derived from solid tumors. The expression value of the cell proliferation gene set has a significant positive correlation, and the proliferation activity of cancer cell lines derived from solid tumors is predicted by the expression level of the cell proliferation gene set;

   2) Establish a prediction model for cell proliferation activity

   a) The single cells in the Tabula Muris database are classified into 81 types according to cell types, and the gene expression values of various types of cells are obtained as above;

   b) Perform hierarchical cluster analysis on 81 cell types using the expression values of the above 87 genes, and cluster the cell types into 2-3 categories through cluster analysis;

   c) For each of the 81 cell types, calculate the expression value of the cell proliferation gene set, and obtain the expression value of 87 genes in the cell proliferation gene set in each cell type. For a certain cell type i, for a certain cell type The gene expression value X _{ji of} a gene j, enumerate the expression vector of its 87 genes as {X _1i ,X _2i ,...,X _87i }, and then calculate the expression value of the cell proliferation gene set as the middle of the 87 gene expression values value;

   d) According to the results of cluster analysis, 81 cell types are grouped into 2-3 different cell type groups, for each cell type group, the expression value vector of its cell proliferation gene set is obtained, and the different cell type groups are compared. The expression value of the cell proliferation gene set, with P<0.05 as the threshold, judge whether the expression value of the cell proliferation gene set of a certain cell type group is significantly higher than that of other cell type groups, so as to confirm the cell type with high expression of cell proliferation gene and low expression The cell types of cell proliferation genes can be used to evaluate the proliferation activity of 81 cell types.
2. The model for predicting cell proliferation activity using 87 genes as biomarkers according to claim 1, wherein the 87 genes are: ANLN, ARHGAP11A, ASF1B, ATAD2, AURKA, AURKB, BIRC5, BRCA2, BUB1 , BUB1B, CCNA2, CCNB1, CCNB2, CDC20, CDC45, CDCA2, CDCA5, CDCA8, CDK1, CDT1, CENPA, CENPE, CENPF, CENPH, CENPK, CENPM, CENPW, CEP55, CKAP2, CKAP2L, CLSPN, DBF4, DLGAP5, ECT2 , ESCO2, FEN1, FOXM1, HIRIP3, HIST1H2AG, HMMR, KIF11, KIF15, KIF20A, KIF20B, KIF23, KIFC1, LMNB1, LMNB2, LRWD1, MAD2L1, MCM2, MKI67, NCAPG, NCAPG2, NCAPH, NDC80, NEIL3, NUDT1, NUDT1 , NUSAP1, PBK, PKMYT1, PLK1, PLK4, PRC1, RACGAP1, RAD51, RCC1, RRM2, SHCBP1, SKA1, SMC2, SNRNP25, SPC24, SPC25, SYCE2, TACC3, TK1, TOP2A, TPX2, TRIM59, TRIP13, TYMS, UBE2C , UBE2T, UHRF1 and VRK1.
3. The model for predicting cell proliferation activity using 87 genes as biomarkers according to claim 1, characterized in that the Tabula Muris database: https://tabula-muris.ds.czbiohub.org/ ; Cancer Genome Atlas Database: http://cancergenome.nih.gov/ ; GTEx database: https://www.gtexportal.org/; from CCLE database: https://portals.broadinstitute.org/ccle .
4. The model for predicting cell proliferation activity using 87 genes as biomarkers according to claim 1, characterized in that: obtaining part of cell proliferation activity data in the CCLE database in step (2) refers to obtaining part of the cell proliferation activity data in the CCLE database Cell doubling time proliferation activity data, using T test to compare the expression values of cell proliferation gene sets of different cell types are using T test.

Images