KR20200038660A - Method for selecting biomarker and method for providing information for diagnosis of cancer using thereof - Google Patents

Abstract

Provided is a method for selecting a biomarker including a step of obtaining single cell RNA sequencing data. By identifying cell type specific alternative polyadenylation in a single cell, cell types based on 3′UTR length changes can be defined with gene expression of various cell types.

Description

Method for selecting biomarker and method for providing information for diagnosis of cancer using thereof

The present invention relates to a method of selecting a biomarker using single cell RNA sequencing data and a method of providing information for diagnosis of cancer using the panel.

Alterative polyadenylation (APA) in the 3 'untranslated region (UTR) is an important post-mortem that regulates gene expression by enriching transcription and affecting cell localization and interaction with microRNA. -It is a post-transcriptional mechanism. Recent studies have revealed that changes in 3'UTR length are closely related to the immune response and regulation of proliferation as well as cell differentiation during cancer growth. The use of 3'UTR shortened through APA is the most common worldwide in cancer (91%). Its biological importance is widely accepted, but its clinical application as a prognostic biomarker or therapeutic target has not been fully evaluated. Thus, understanding expression regulation through APA for various cell types can provide new insights into cancer therapeutics.

One aspect is obtaining RNA sequencing data of a patient-derived cell; Estimating alternative polyadenylation (APA) of individual genes from the obtained RNA sequencing data; Calculating an index indicating the expression level of individual genes; And measuring the correlation between the calculated index and the replacement polyadenylation.

Another aspect is to provide a biomarker panel selected by the above method.

Another aspect is obtaining RNA sequencing data of a patient-derived cell; Estimating alternative polyadenylation (APA) of individual genes from the obtained RNA sequencing data; Calculating an index indicating the expression level of individual genes; And measuring the correlation between the calculated index and the replacement polyadenylation.

One aspect is obtaining RNA sequencing data of a patient-derived cell; Estimating alternative polyadenylation (APA) of individual genes from the obtained RNA sequencing data; Calculating an index indicating the expression level of individual genes; And measuring the correlation between the calculated index and the replacement polyadenylation.

Conventional biomarker panel studies using bulk tissue transcript data have a problem that the configuration of biomarker panels according to cell types is limited due to difficulty in reflecting microenvironment and heterogeneity characteristics in tissues. However, the method according to one aspect uses a single-cell transcript data to analyze the association of gene expression with alternative RNA polyadenylation, a transcription control mechanism, and is specific for immune and tumor cell-specific and existing gene expression present in cancer. A new breast cancer biomarker panel that could not be found in the panel composition of can be constructed. In one embodiment, the gene constituting the biomarker panel is closely related to RNA polyadenylation in a cancer patient population, and there is a population of cells having a significantly different expression level of a specific gene in a patient population. It means that changes in gene expression regulation have occurred. In particular, when the expression level of the gene is markedly increased and the survival result of the gene cluster is statistically significant and different, the change in the expression level of this gene in epithelial cancer is closely related to the prognosis of the cancer. can see. The upper and lower 25th percentiles of gene expression were classified into high and low expression groups using cutoff.

The method according to one embodiment comprises obtaining RNA sequencing data of a patient-derived cell. The cells may be single cells. For example, the tumor cells may be B lymphocytes, T lymphocytes, myeloid cells or stromal cells. In addition, the data may be an RNA full-length transcript.

The method according to one embodiment comprises estimating the replacement polyadenylation of an individual gene from the obtained RNA sequencing data. The replacement polyadenylation can be estimated using DaPars or Roar. Specifically, alternative polyadenylation can be estimated by measuring the relative changes in the amount of short isoform and long isoform of a gene from RNA-seq data. In addition, the replacement polyadenylation can be estimated by changing the length of 3'UTR for individual genes. In addition, patient-derived cells can be isolated from a variety of tumors, for example, breast cancer, glioblastoma, renal cell carcinoma, adenocarcinoma, mucus cancer, phantom cell, squamous cell carcinoma, adenosquamous cell carcinoma, small cell carcinoma, chorionic cancer, undifferentiated cancer And other epithelial cancers. Alternative polyadenylation of individual genes can be estimated by adjusting the replacement polyadenylation according to the cell type.

The method according to one embodiment includes calculating an index indicating the expression level of an individual gene. The expression level of individual genes is, for example, microarray, multiplex polymerase chain reaction (PCR), quantitative RT-PCR (quantitative reverse transcription polymerase chain reaction), transcriptome analysis using a tiling array , Can be measured using short read sequencing. In addition, the calculation of the index indicating the expression level of the individual gene is within samples or between samples, such as Fragments Per Kilobase of transcript per Million (FPKM), Reads Per Kilobase Million (RPKM), Transcripts Per Kilobase Million (TPM), Quantile-nomrmalization, etc. Normalization can be used. Specifically, read count of an individual gene may be secured through read alignment using a STAR method and read quantification using an RSEM method from the obtained RNA sequencing data. The read count can be normalized to a transcript per million (TPM) level for comparison between cells / samples, and converted to log2 level (log2 (TPM + 1)) to quantify the degree of gene expression.

The method according to one embodiment includes measuring the correlation between the calculated index and the replacement polyadenylation. The calculated index can be expressed as the number of normalized RSEMs in the top 4 quartiles, and can be quantified using transcript data produced from single cells derived from cancer patients. The correlation can be measured by calculating a correlation coefficient. For example, the correlation coefficient can be calculated using Pearson's correlation coefficient.

As used herein, the term "RESM number" refers to an expression estimate derived from RNA-seq data using a standardized transcript quantification technique (RSEM). At this time, the quartile normalization is one of the normalization methods, and the TCGA data used for the survival rate analysis means that the degree of gene expression is quantified by the corresponding standardization method.

In addition, the method may further include calculating a statistical significance level representing the difference in survival rates between the classified patient groups, and selecting genes based on the calculated statistical significance level. Survival analysis refers to a statistical technique for determining whether survival rate differences between predicted prognostic groups can be classified. The survival analysis can use various methods commonly used in the art. For example, the single-variable Kaplan-meier method, the log-rank test, or the multivariate analytical cox proportional-hazards regression model. Can be used. In addition, the method may be for diagnosing early cancer. The method shows a differential pattern for 3'UTR shortening depending on the cancer type, and thus may have tumor specificity.

Another aspect provides a biomarker panel selected by the above method. The biomarker panel is a single cell transcript data that considers alternative polyadenylation and expression patterns of individual genes and is closely related to the survival rate of early cancer patients. Can be configured. In one embodiment, the panel may include an agent that measures the level of the biomarker selected by the method. For example, it may include an agent that measures the level of two or more biomarkers selected by the above method.

The agent measuring the level of the biomarker may be a primer pair, a probe or an antisense nucleotide. Specifically, it may be an agent for measuring the mRNA level of the biomarker gene, and may be a primer pair, probe or antisense nucleotide specifically binding to the gene. In addition, the agent for measuring the level of the biomarker may be an antibody. The antibody may be a monoclonal antibody, for example, a monoclonal antibody that specifically binds to any of the biomarkers.

Another aspect is obtaining RNA sequencing data from a patient; Estimating alternative polyadenylation (APA) of individual genes from the obtained RNA sequencing data; Calculating an index indicating the expression level of individual genes; And measuring a correlation between the calculated index and the replacement polyadenylation.

The method according to one embodiment comprises obtaining RNA sequencing data of a patient-derived cell. The details of obtaining the data are as described above.

The method according to one embodiment includes estimating polyadenylation of individual genes from the obtained RNA sequencing data. The details of the estimation are as described above.

The method according to one embodiment includes calculating an index indicating the expression level of an individual gene. The details of the calculation are as described above.

The method according to one embodiment includes measuring the correlation between the calculated index and the replacement polyadenylation. Specifically, when the expression level of the individual gene of the individual and the replacement polyadenylation increase compared to the control group, it can be judged as cancer. At this time, the expression level of the individual gene is calculated as the expression level of the individual gene and can be expressed by an index. In addition, the cancer may be an early cancer.

The method according to one aspect can provide a computational approach to exploring transcription signatures associated with regulating gene expression and alternative polyadenylation of 3'UTR using single cell RNA sequencing data. In addition, by identifying cell type specific alternative polyadenylation in a single cell, it is possible to define cell types based on 3'UTR length changes with gene expression of various cell types. In particular, an alternative polyadenylation signal that is immune specific in breast cancer can be used as a prognostic indicator of early breast cancer.

1 shows APA pattern differences between tumor and non-tumor cells. 1A shows the signal of APA (PDUI calculated by DaPars) for all genes, and FIGS. 1B and 1C show hierarchical clustering of genes and gene sets for individual cells derived from breast cancer patients, respectively. 1D shows Unsupervised tSNE for gene set-level APA separating tumor cells and non-tumor cells from breast cancer. Individual cells are sampled and match the coloration of FIGS. 1B and 1C.
Figure 2 shows cell type-specific functional categories related to 3'UTR shortening and overexpression. FIG. 2A shows the Venn diagram for a set of 1,176 genes specific to 5 cell types and FIG. 2B shows the results of a network-based functional analysis of selected gene sets.
Figure 3 shows the expression change of the gene related to 3'UTR shortening and proliferation according to cell type. The cross ratio and its significance are marked with a specific cell type in the upper right of the plot, and the line of alignment is colored to identify the cell type that matches.
4 shows the heterogeneity of APA regulation between cancer types. Figure 4a shows hierarchical clustering for gene set-level APA estimation for 598 single cells from 3 cancer types, Figure 4b for gene set-level APA separating tumor cells with cancer type Unsupervised tSNE and FIG. 4C shows the APA map of the top 10 gene set in each cancer type (p <0.01). Cancer types and colors for patients are shown in FIG. 4A.
5 shows cell type specific genes with clinical relevance. FIG. 5A shows the Venn diagram for 53 gene hits specific to 5 cell types, the genes in red are typical markers in the results of Roar and the genes in blue are de novo markers in the public APA database. FIG. 5B shows Kaplan-Meier survival curves for SET, HSP90AA1, YWHAZ, and DDX5 in TCGA BRCA patients. Tumor samples show 'high' and 'low' (25th and 75th percentiles) for the expression signal of each gene. Su) The group was annotated. The p-value is determined for log-rank testing.

Hereinafter, preferred embodiments are provided to help understanding of the present invention. However, the following examples are only provided to more easily understand the present invention, and the contents of the present invention are not limited by the following examples.

[Example]

Example 1 Data Collection

Raw RNA-seq data for 515 cells from 11 breast cancer patients and 30 cells from kidney cancer patients were obtained using accession codes GSE75688 and GSE73122 from the NCBI Gene Expression Omnibus database. In addition, Raw RNA-seq data for 288 cells were downloaded from 3 glioblastoma patients using the accession code EGAS00001001880 from the European Genome-phenome Archive (EGA). This data was generated in the C1 Single-Cell Auto Prep System (100-5760, Fluidigm, San Francisco, CA, USA) as a full length transcript. To evaluate the use of 3'UTR length, a .bam file was created with the Roar method input using the 2-pass mode (basic parameter) of STAR_2.4.0b. In addition, using the 'genomCoverageBed' command (BEDtools v2.17.0), a .bedgraph file was created from the .bam file with DaPars method input. Thereafter, 34,942 genes were extracted for expression analysis of breast cancer, and the expression values of these genes were assumed in at least one cell. The relative expression of each gene is expressed in transcript per million (TPM) using RSEM v1.2.17 (basic parameter). As a control, Raw RNA-seq data of normal breast, brain, and kidney tissue from the Body Map 2.0 project can be used in ArrayExpress (http://www.ebi.ac.uk/arrayexpress Query ID: E-MTAB- 513.). Sequential methods for read alignments and quantification were applied in the same manner as the single cell RNA-seq data process. For survival analysis, RNA-seq and clinical data of patients' Invasive Carcinoma (BRCA) samples were obtained from The Cancer Genome Atlas (TCGA). In this RNA-seq data (level 3), there are 1,073 tumors (updated in 2017), and the expression of each gene is expressed by the number of top four quintile normalized RNA-Seq by Expectation Maximization (RSEM). Subtypes of BRCA tumors were predicted using the R package 'genefu'.

Example 2. Confirmation of cellular heterogeneity of 3'UTR length change

Using full-length single cell RNA sequencing data generated from tumor tissue in breast cancer patients, we predict the change in 3'UTR length in five major cell types, including tumor cells, B lymphocytes, T lymphocytes, and stromal cells. Compared. Two complementary methods were applied to determine the shortening and extension of the 3'UTR. Alternative polyadenylation (APA) due to length change of 3'UTR was estimated using DaPars (basic parameter) and Roar. DaPars scans all regions of the 3'UTR in the reference genome (hg19) to detect new APA sites, while Roar focuses on the known APA sites of the 3'UTR to improve sensitivity, mainly estimated by DaPars 3 'We used the UTR switching results. Specifically, .gtf files generated from public APA databases of PolyA_DB2 and APASdb were used using single cell and bulk RNA-seq samples.

1 shows the pattern difference of APA between tumor and non-tumor cells. 1A and 1B, in all cells derived from tumor tissue of breast cancer patients, 3'UTR shortening was predominant. In addition, to increase the resolution of patterning, a single sample GSEA (ssGSEA) was used to compare the overall APA pattern at the gene set level. As a result, as shown in Figs. 1C and 1D, it was confirmed that 3'UTR shortening was not affected by the cancer subtype and sample arrangement.

Example 3. Construction of the gene set

APA measurement of 'change in Percentage of Distal polyA site Usage Index' (PDUI, by DaPars) and 'Ratio of A Ratio' (roar, by Roar), respectively Used as input data. Gene expression quantified as transcript per million (TPM) converted log2 to plus 1. In addition, in order to evaluate APA regulation and gene expression based on pathway activation, all gene symbols were matched with EntrezID, and then ssGSEA (using the 'GSVA' option of the R package) was applied to obtain an enrichment score per gene set. It was calculated. A total of 5,917 gene ontology (GO) terms were collected for ssGSEA published in the gene set database, MSigDB v6.0.

Example 4. Selection of specific signatures by cell type

Gene expression and APA levels were compared for five major cell types, including tumor cells, B lymphocytes, T lymphocytes, and stromal cells. Specifically, two measures were used to determine the association between the use of 3'UTR and gene expression regulation. Correlation between 3'UTR length change and gene expression in each cell type was calculated using the Pearson's correlation coefficient (PCC) as a concentration score scale for the gene set. To determine the statistical significance of PCC, p-values were calculated based on Fisher's Z transformation. The odds ratio (OR) was calculated by quantifying the cell population for the specificity of 3'UTR shortening and overexpression for the cell type. All single cells were classified into 4 groups using the change of 3 'UTR length and the expression level of the gene, and the cross ratio was calculated;

(a) Cells showing 3'UTR shortening and overexpression for specific cell types

(b) Cells showing 3'UTR shortening and overexpression to other cells

(c) Cells that do not show 3'UTR shortening and overexpression for specific cell types

(d) Cells that do not show 3'UTR shortening and overexpression in other cells.

APA levels and median gene expression were used as criteria for classifying the groups to determine whether individual cells exhibit 3'UTR shortening and overexpression for each gene set and gene. The Fisher exact test was then used to determine the statistical significance of the matches between individual query pairs. For the selection of hits, cutoffs were applied for PCC> 0 (p value <0.05) and OR> 2 (p value <0.01).

Immune and stromal cell specific gene sets have formed functional categories of immune and inflammatory responses. In particular, during cell proliferation, clustering patterns of gene sets specific to four cell types except stromal cells appeared. The general shortening of 3'UTR length is closely related to the state of cell proliferation and dedifferentiation. It was also confirmed that 3'UTR shortening and gene expression for a set of genes classified as cell proliferation each exhibited a limited correlation pattern associated with its cell type (FIG. 3). These results suggest that APAs involved in regulating expression are highly dependent on unique cell lineages. Thus, the understanding of APA in single cell analysis is useful for recognizing differences in transcriptional regulation between various cells in cancer.

Example 5. Selection of tumor specific gene sets

To compare the prevalence of APA mutations in cancer tissues, public full-length single cell RNA sequencing data generated from glioblastoma and renal cell carcinoma tissues was obtained, and 280 tumor cell APA-prediction data were used in tumor tissues of breast cancer patients. Delta and t-test p values were calculated using APA prediction data modified with the gene set to select a gene set showing significant 3'UTR switching for each cancer type. Afterwards, a list of the most differential gene sets (top 10 hits) in each cancer was prepared. The delta of the gene set is expressed as the difference between the average of a single cell for a given tumor type and other tumors, and the significance (p value) was calculated by bi-t statistics.

As a result, as shown in Figures 4a and 4b, it was confirmed that the tumor tissue is clearly distinguished through clustering based on 3'UTR length change at the gene set level. In addition, the gene set contributing to the tumor-specific cluster through differential APA regulation was further investigated. As a result, it was confirmed that 739, 898, and 731 gene sets were significantly switched in breast cancer, glioblastoma, and kidney cell cancer, respectively (p <0.01). In addition, as shown in Figure 4c, it was possible to confirm the differential pattern for 3'UTR shortening in each cancer.

Example 6. Construction of a clustering gene set based on a network

Cytoscape (v3.5.1) was used to construct clusters based on the biological function of the gene set with interactive network graphics. The edge of the network refers to the Jaccard index for the number of GO terms shared between two sets of genes. The distance between the nodes (gene set) was defined using force-directed layout. The network was expressed by selecting a set of 906 genes with sufficient functional association (Jaccard index> 0.5 and> 9 gene sets per cluster).

Example 7. Survival Analysis-Clinical relevance of APA-related marker genes

To reveal the association between APA regulation and gene markers, single cell sequencing data was re-analyzed at the gene level in breast cancer patients, using shortened 3'UTRs and finding 53 genes specifically overexpressed in each cell type (Fig. 5a). As a result, as shown in FIG. 5A, in accordance with the result of the gene set level, it was confirmed that most of the hit genes are classified by cell type except for the 4 overlapping genes for the immune cell type. Next, in order to confirm the clinical effect of these genes, Kaplan-Meier survival analysis was performed and TCGA RNA sequencing data was used to investigate the relationship between expression change and survival in breast cancer patients. Tumor samples were divided into two groups, 'high' and 'low' (25th and 75th percentiles) according to the expression signal of each target gene. In this analysis, using all 1,091 BRCA samples, 16 out of 53 genes showed significant survival (p <0.01, 10 genes showed p <0.05) between the two groups (FIG. 5B, Table 1). Reference). In the package 'OIsurv', a survival curve was drawn using the Kaplan-Meier formula. By changing the sample window for each cancer stage, it was confirmed that in the early stage tumor, a total of 17 gene expression levels significantly (p <0.1, 11 genes showed p <0.05) genetic effects on survival. In contrast, in the late stage tumor, when the cutoff p-value was 0.1 or 0.5, it was confirmed that only 7 genes or 2 genes affected survival. However, there was no difference in the survival rate for the molecular subtype of breast cancer. In addition, a multivariate Cox regression analysis was performed to investigate the relative risk in the R package 'survival'. A regression model was constructed considering 10 factors such as age, race, stage, tumor weight, presence of ER / PR / Her2, drug / radiation therapy indication, and expression class of each gene. Of the 16 genes that passed the p value of the Kaplan-Meier survival analysis in the tissue sample, it was confirmed that the expression level of 11 genes was an independent (p <0.1) factor affecting the survival of BRCA patients (Table 1).

gene Cell type Kaplan- Meier  (p-value) Cox regression analysis method (Cox-
Regression)
(p-value) step Sub type gun
(1,073) Early
(792) review
(267) Basal
(222) Her2
(90) Luminal  A
(329) Luminal  B
(432) gun
(1,073) ALDOA tumor 0.071 0.207 0.489 0.141 0.427 0.118 0.891 0.042 CHCHD2 tumor 0.022 0.155 0.151 0.861 0.261 0.388 0.07 0.142 GGCT tumor 0.018 0.131 0.077 0.495 0.008 0.812 0.035 0.1 H3F3A tumor 0.857 0.455 0.962 0.323 0.421 0.285 0.169 0.959 HSP90AB1 tumor 0.034 0.012 0.624 0.465 0.807 0.075 0.104 0.019 MYL12B tumor 0.459 0.838 0.224 0.827 0.601 0.598 0.72 0.64 NME1 tumor 0.877 0.96 0.865 0.779 0.957 0.814 0.704 0.77 NR4A1 tumor 0.235 0.25 0.379 0.414 0.193 0.588 0.862 0.73 PABPC1 tumor 0.093 0.196 0.085 0.513 0.872 0.546 0.147 0.061 SET tumor 0.086 0.019 0.847 0.896 0.284 0.412 0.293 0.002 SOD1 tumor 0.604 0.523 0.845 0.832 0.516 0.589 0.098 0.845 TSTA3 tumor 0.136 0.41 0.153 0.016 0.605 0.327 0.327 0.135 ZNF706 tumor 0.015 0.124 0.21 0.4 0.478 0.832 0.003 0.035 CFL1 T cell 0.389 0.741 0.197 0.896 0.162 0.49 0.221 0.697 CLIC1 T cell 0.096 0.842 0.0496 0.656 0.074 0.582 0.073 0.345 CTSB T cell 0.113 0.243 0.539 0.566 0.929 0.554 0.483 0.013 GABARAP T cell 0.809 0.559 0.216 0.727 0.925 0.462 0.855 0.82 HSP90AA1 T cell 0.001 0 0.917 0.777 0.138 0.276 0.026 0 PSMC2 T cell 0.09 0.477 0.113 0.232 0.117 0.094 0.496 0.418 RPL23A T cell 0.119 0.477 0.149 0.459 0.674 0.343 0.148 0.58 SH3BGRL3 T cell 0.223 0.043 0.05 0.635 0.044 0.588 0.933 0.484 TMEM126B T cell 0.179 0.083 0.685 0.894 0.167 0.505 0.546 0.036 WAC T cell 0.017 0.027 0.897 0.861 0.152 0.099 0.268 0.003 TMEM59 Stroma cells
(Stromal cell) 0.885 0.704 0.983 0.96 0.195 0.022 0.623 0.15 TSC22D1 Stroma cells 0.177 0.187 0.727 0.02 0.094 0.608 0.261 0.201 B2M Bone marrow & T cells 0.164 0.193 0.969 0.074 0.633 0.767 0.823 0.408 RHOA Bone marrow & T cells 0.417 0.022 0.178 0.221 0.243 0.178 0.409 0.118 ATP6V0E1 marrow 0.562 0.153 0.057 0.206 0.629 0.655 0.257 0.17 HLA-A marrow 0.252 0.074 0.143 0.023 0.279 0.427 0.946 0.194 SEP15 marrow 0.35 0.8 0.744 0.02 0.186 0.926 0.557 0.575 DDX5 B cells &
T cell 0.632 0.09993 0.224 0.96 0.646 0.029 0.853 0.061 LAP3 B cells &
T cell 0.344 0.809 0.124 0.154 0.78 0.667 0.768 0.375 ACTB B cell 0.563 0.263 0.004 0.702 0.749 0.142 0.125 0.138 ACTG1 B cell 0.52 0.417 0.391 0.584 0.543 0.287 0.166 0.842 ARPC2 B cell 0.11 0.059 0.536 0.265 0.423 0.525 0.056 0.031 BRK1 B cell 0.09 0.082 0.537 0.139 0.013 0.464 0.053 0.112 CLK1 B cell 0.687 0.267 0.816 0.869 0.329 0.007 0.25 0.231 CSNK1A1 B cell 0.006 0.008 0.825 0.716 0.061 0.014 0.037 0 EIF2S1 B cell 0.2 0.131 0.899 0.581 0.499 0.689 0.168 0.05 EIF4A2 B cell 0.675 0.12 0.326 0.766 0.31 0.177 0.631 0.066 FKBP1A B cell 0.142 0.344 0.068 0.583 0.519 0.66 0.062 0.04 GPBP1L1 B cell 0.691 0.27 0.245 0.471 0.396 0.242 0.459 0.253 GSPT1 B cell 0.232 0.022 0.247 0.665 0.23 0.277 0.358 0.007 HNRNPA1 B cell 0.401 0.718 0.19 0.572 0.45 0.092 0.367 0.751 HNRNPK B cell 0.558 0.077 0.153 0.034 0.881 0.273 0.729 0.056 LTV1 B cell 0.025 0.001 0.721 0.9 0.824 0.054 0.03 0.003 MMADHC B cell 0.049 0.002 0.447 0.249 0.248 0.066 0.092 0.004 PFN1 B cell 0.589 0.988 0.434 0.912 0.192 0.488 0.026 0.811 SNAP23 B cell 0.985 0.374 0.209 0.518 0.233 0.504 0.674 0.108 SUMO2 B cell 0.866 0.227 0.937 0.02 0.467 0.169 0.198 0.317 TAF9 B cell 0.511 0.507 0.857 0.766 0.695 0.192 0.595 0.016 UBB B cell 0.334 0.152 0.399 0.281 0.4 0.337 0.164 0.624 YWHAZ B cell 0.006 0.007 0.263 0.764 0.337 0.738 0.224 0.001

The above description of the present invention is for illustration only, and a person having ordinary knowledge in the technical field to which the present invention pertains can understand that it can be easily modified into other specific forms without changing the technical spirit or essential features of the present invention. will be. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive.

Claims (15)

Obtaining RNA sequencing data of a patient-derived cell;
Estimating alternative polyadenylation (APA) of individual genes from the obtained RNA sequencing data;
Calculating an index indicating the expression level of individual genes; And
A method of screening for a biomarker comprising measuring the correlation between the calculated index and alternative polyadenylation. The method of claim 1, further comprising calculating a statistical significance level representing a difference in survival rates between the classified patient groups, and selecting a gene based on the calculated statistical significance level. The method of claim 1, which is single cell RNA sequencing data. The method according to claim 3, wherein the single cell is selected from the group consisting of tumor cells, B lymphocytes, T lymphocytes, myeloid cells and stromal cells. The method of claim 1, wherein the data is an RNA full-length transcript. The method of claim 1, wherein the replacement polyadenylation is estimated using DaPars or Roar. The method of claim 1, wherein the replacement polyadenylation is estimated by a change in the length of 3′UTR of an individual gene. The method according to claim 1, wherein the cell is selected from the group consisting of breast cancer, glioblastoma, renal cell carcinoma, adenocarcinoma, mucous cancer, leukocytosis, squamous cell carcinoma, squamous cell carcinoma, small cell carcinoma, chorionic cancer, and undifferentiated cancer. The method of claim 1, wherein alternative polyadenylation is regulated according to cell type. The method of claim 9, wherein the cell is selected from the group consisting of tumor cells, B lymphocytes, T lymphocytes, myeloid cells and stromal cells. The method of claim 1 having tumor specificity. Biomarker panel selected by claim 1. Obtaining RNA sequencing data of a patient-derived cell;
Estimating alternative polyadenylation (APA) of individual genes from the obtained RNA sequencing data;
Calculating an index indicating the expression level of individual genes; And
Method for providing information for diagnosis of cancer, comprising the step of measuring the correlation between the calculated index and alternative polyadenylation. The method of claim 13, further comprising the step of judging cancer if the expression level of the individual gene of the individual and the replacement polyadenylation increase compared to the control group. The method of claim 13 for diagnosing early cancer.

Images