基因于肿瘤TNM分期相关性（包括四个表卡方检验+annovar检

第一部分

step1

phe_table<- table(phe$Sex,phe$`N stage`)
phe_table
chisq.test(phe_table)
fisher.test(phe_table)

##### 卡方检验 N 相关
rm(list = ls())
load(file = './Rdata/!step0_cleaned_cancer_sample.Rdata')
# gene_name<- c('GPD1L')
gene_name<- c('GPD1L')
match(colnames(exprSet),phe$GSM_ID)
phe$N_stage<- factor(phe$`N stage`)
phe$N_stage<- ifelse(phe$N_stage=='N0','N+','N0')
phe$group=ifelse(exprSet[gene_name,]>quantile(exprSet[gene_name,],0.5),'high','low')
phe_table<- table(phe$N_stage,phe$group)
phe_table
chisq.test(phe_table)
fisher.test(phe_table)

第二部分

#######################################  整理数据
rm(list = ls())
load(file = './Rdata/!step0_cleaned_cancer_sample.Rdata')

exprSet[1:4,1:4]

gene_name<- c('SPP1')
library(ggpubr)
library(ggplot2)
library(reshape2)
gene_expression<- as.data.frame(exprSet[gene_name,])
names(gene_expression)<- gene_name
match(colnames(exprSet),phe$GSM_ID)


# gene_expression$group<- factor(phe$`T stage`)
# gene_expression$group<- ifelse(gene_expression$group== 'T1'|gene_expression$group=='T2','T1-2','T3-4')


gene_expression$group<- factor(phe$`N stage`)
# gene_expression$group<- ifelse(gene_expression$group!='N0','N+','N0')
# gene_expression$samlple<- rownames(gene_expression)
exprSet_L<- melt(gene_expression,id.vars = c('group'))
names(exprSet_L)[2]<- c('sample')
p=ggplot(exprSet_L,aes(x=sample,y=value,fill=group))+geom_boxplot()+ 
  stat_compare_means(method = "aov",label="p.signif")
print(p)

# ggsave('./figure/ggplot_boxplot_T.pdf',p,width = 40, height = 20, units = "cm")
# ggsave('./figure/ggplot_boxplot_N.pdf',p,width = 40, height = 20, units = "cm")