TCGA数据挖掘入门---任意基因的任意分组比较

任意基因的任意分组比较是生信技能树[生信爆款入门课程]TCGA数据挖掘部分提到的一个重点。为拓展课堂所学知识，现在对其做下练习巩固和总结。

1.加载并查看输入数据

> rm(list=ls())
> load("for_boxplot.Rdata")

数据包含信息如下

image.png

数据解读
exp是tumor-normal都有的表达矩阵，exprSet是只有tumor样本的表达矩阵。meta是临床信息表格，Group是tumor-normal分组信息。mut是突变信息，由maf文件读取并取子集得到。

2.比较任意miRNA在tumor和normal样本中的表达量

>以hsa-mir-143为例
> table(Group)
Group
normal  tumor 
    71    522 
> library(ggstatsplot)
> dat = data.frame(gene = exp["hsa-mir-143",],
+                  group = Group)
> ggbetweenstats(data = dat, x = group,  y = gene,title = "hsa-mir-143")
>

image.png

3.任意miRNA在任意两个分组中的表达量对比

只要是可以根据临床信息查到或得到的分组，例如生死、人种、阶段，都可以拿来做分组，需要注意的是需要调整样本顺序，使一一对应。

按照生死、人种、分期分组查看
 table(meta$patient.vital_status)

alive  dead 
  358   158 
> table(meta$patient.stage_event.pathologic_stage)

  i  ii iii  iv 
254  55 124  83 
> table(meta$patient.race)

                    asian black or african american                     white 
                        8                        56                       445 

> dat = data.frame(gene = exprSet["hsa-mir-143",],
+                  vital_status = meta$patient.vital_status,
+                  stage = meta$patient.stage_event.pathologic_stage,
+                  race = meta$patient.race)
> p1 = ggbetweenstats(data = dat, x = vital_status,  y = gene,title = "hsa-mir-143")
p1

image.png

> p2 = ggbetweenstats(data = dat, x = stage,  y = gene,title = "hsa-mir-143")
> p2

image.png

> p3 = ggbetweenstats(data = dat, x = race,  y = gene,title = "hsa-mir-143")
> p3

image.png

4.根据某个基因是否突变分组比较某miRNA的表达量

> dim(exprSet)
[1] 552 516
> head(mut)
   Hugo_Symbol Chromosome Start_Position         Tumor_Sample_Barcode     t_vaf
1:    HNRNPCL2       chr1       13115853 TCGA-G6-A8L7-01A-11D-A36X-10 0.2148148
2:       ERMAP       chr1       42842993 TCGA-G6-A8L7-01A-11D-A36X-10 0.1650165
3:        FAAH       chr1       46394349 TCGA-G6-A8L7-01A-11D-A36X-10 0.3114754
4:       EPS15       chr1       51448116 TCGA-G6-A8L7-01A-11D-A36X-10 0.1677852
5:      HMGCS2       chr1      119764248 TCGA-G6-A8L7-01A-11D-A36X-10 0.2539683
6:      NOS1AP       chr1      162367063 TCGA-G6-A8L7-01A-11D-A36X-10 0.2098765
              pos
1:  chr1:13115853
2:  chr1:42842993
3:  chr1:46394349
4:  chr1:51448116
5: chr1:119764248
6: chr1:162367063
> library(stringr)
> length(unique(str_sub(mut$Tumor_Sample_Barcode,1,12)))
[1] 336
> k = str_sub(colnames(exprSet),1,12) %in% unique(str_sub(mut$Tumor_Sample_Barcode,1,12));table(k)
k
FALSE  TRUE 
  185   331 
> 
> #516个样本中,有331个有突变信息记录,将这些样本对应的表达矩阵取出来。
> expm = exprSet[,k]
> 
> VHL_mut=str_sub(as.character(
+   as.data.frame( mut[mut$Hugo_Symbol=='VHL','Tumor_Sample_Barcode'])[,1] ),
+   1,12)
> 
> library(dplyr)
> VHL_mut = mut  %>% 
+   filter(Hugo_Symbol=='VHL')  %>%   
+   as.data.frame()  %>% 
+   pull(Tumor_Sample_Barcode)   %>%  
+   as.character()   %>%   
+   str_sub(1,12)
> 
> #false 是未突变样本，true是突变样本
> 
> tail(rownames(expm))
[1] "hsa-mir-944" "hsa-mir-95"  "hsa-mir-96"  "hsa-mir-98"  "hsa-mir-99a"
[6] "hsa-mir-99b"
> dat=data.frame(gene=expm['hsa-mir-98',],
+                mut= str_sub(colnames(expm),1,12) %in% VHL_mut)
> 
> ggbetweenstats(data = dat, x = mut,  y = gene)

image.png

5.计算每个基因的p值

看看差异是不是显著。

> res.aov <- t.test(gene ~ as.factor(mut), data = dat)
> res.aov$p.value
[1] 0.9086252

最后编辑于：2021.02.07 11:30:25