Survival curve for TCGA

 cgdsr

library(cgdsr)

library(DT)

mycgds = CGDS("http://www.cbioportal.org/")

test(mycgds)

### all cancer studies

all_tcga_studies <- getCancerStudies(mycgds)

datatable(all_tcga_studies[,c(1,3)])

### get the lggGBM genetic profiles

lgg_genetic <- getGeneticProfiles(mycgds,'lgg_tcga_pan_can_atlas_2018')[,c(1:3)]

datatable(lgg_genetic)

mygenetic <- "lgg_tcga_pan_can_atlas_2018_rna_seq_v2_mrna"

### get the case of lgg

lgg_case <- getCaseLists(mycgds,'lgg_tcga')[,c(1:3)]

datatable(lgg_case)

mycase <- "lgg_tcga_all"

df = getProfileData(mycgds,"TP53",mygenetic,mycase)

df <- na.omit(df)

cldt <- getClinicalData(mycgds,"lgg_tcga_all")

group <- vector(length = 419L)

df_all_data$PERSON_NEOPLASM_CANCER_STATUS <- revalue(factor(df_all_data$PERSON_NEOPLASM_CANCER_STATUS),c(With Tumor = 1,Tumor Free = 0))

gene_name <- "ASCL1"

time_status <- select(df_all_data,c(DAYS_LAST_FOLLOWUP,PERSON_NEOPLASM_CANCER_STATUS,gene_name))

time_status$group <- sapply(time_status[gene_name], function(x){ifelse(x-median(x) <0,"low","high")})

time_status <- time_status[time_status$PERSON_NEOPLASM_CANCER_STATUS !="",]

time_status$PERSON_NEOPLASM_CANCER_STATUS <-revalue(time_status$PERSON_NEOPLASM_CANCER_STATUS,c("Tumor Free" =0,"With Tumor" =1))

time_status$PERSON_NEOPLASM_CANCER_STATUS <- time_status$PERSON_NEOPLASM_CANCER_STATUS  %>% droplevels()

time_status$PERSON_NEOPLASM_CANCER_STATUS <- as.integer(time_status$PERSON_NEOPLASM_CANCER_STATUS)

mysurvfit <-  survfit(Surv(time_status$DAYS_LAST_FOLLOWUP, time_status$PERSON_NEOPLASM_CANCER_STATUS) ~time_status$group,data = time_status)

mysurvfit2 <-  survfit(Surv(time_status$DAYS_LAST_FOLLOWUP, as.integer(time_status$PERSON_NEOPLASM_CANCER_STATUS)) ~time_status$group,data = time_status)

plot(mysurvfit2)

ggsurvplot(mysurvfit2,pval = T)

###### CONSTRUCT A LOOP

everyfit <- function(gene_name = "ASCL1"){

  time_status <- select(df_all_data,c(DAYS_LAST_FOLLOWUP,PERSON_NEOPLASM_CANCER_STATUS,gene_name))

  time_status$group <- sapply(time_status[gene_name], function(x){ifelse(x-median(x) <0,"low","high")})

  time_status <- time_status[time_status$PERSON_NEOPLASM_CANCER_STATUS !="",]

  time_status$PERSON_NEOPLASM_CANCER_STATUS <-revalue(time_status$PERSON_NEOPLASM_CANCER_STATUS,c("Tumor Free" =0,"With Tumor" =1))

  time_status$PERSON_NEOPLASM_CANCER_STATUS <- time_status$PERSON_NEOPLASM_CANCER_STATUS  %>% droplevels()

  time_status$PERSON_NEOPLASM_CANCER_STATUS <- as.numeric(time_status$PERSON_NEOPLASM_CANCER_STATUS)

  time_status$PERSON_NEOPLASM_CANCER_STATUS[time_status$PERSON_NEOPLASM_CANCER_STATUS ==1] <- 0

  time_status$PERSON_NEOPLASM_CANCER_STATUS[time_status$PERSON_NEOPLASM_CANCER_STATUS ==2] <-1

  thisfit <-  survfit(Surv(time_status$DAYS_LAST_FOLLOWUP, time_status$PERSON_NEOPLASM_CANCER_STATUS) ~time_status$group,data = time_status)

  invisible(thisfit)

}

everyfit("NOTCH1") %>% ggsurvplot(pval = T,conf.int=F)

everyfit("MAP2") %>% ggsurvplot(pval = T)

RTCGA

### load packages

library(RTCGA)

library(RTCGA.clinical)

library(RTCGA.rnaseq)

library(dplyr)

library(DT)

#????????????

infoTCGA <- infoTCGA() #??????????????????????????????????????????????????????????????????????????????

# Create the clinical data

#library(RTCGA.clinical)

clin <- survivalTCGA(GBMLGG.clinical)

clin[1:5,] ## status 0 = alive,1 = dead

##library(RTCGA.mRNA) #???????????????

class(GBMLGG.rnaseq)  #???????????????????????????????????????

dim(GBMLGG.rnaseq)  #??????????????????????????????696????????????20532?????????

GBMLGG.rnaseq[1:5,1:5] ## row for sample col for genes

###read my gene_List

list.files()

library(xlsx)

genelist <- read.xlsx("gene_List.xlsx",1,header = F)

genelist

myvector <- vector(mode = "list",length = length(genelist$X1))

for (i in seq_along(genelist$X2)) {

  myvector[[i]] <- grep(genelist$X2[i],names(GBMLGG.rnaseq),value = T)

}

myvector[[2]] <- myvector[[2]][1]

myvector[[3]] <- myvector[[3]][2]

myvector[[5]] <- myvector[[5]][4]

myvector[[7]] <- myvector[[7]][2]

myvector[[8]] <- myvector[[8]][3]

myvector <- unlist(myvector)

#### select

exprSet <- GBMLGG.rnaseq %>% as_tibble() %>%

  select("bcr_patient_barcode",myvector) %>%

  mutate(bcr_patient_barcode = substr(bcr_patient_barcode,1,12)) %>%

  inner_join(clin,by="bcr_patient_barcode")

colnames(exprSet)[2:9] <- sub("\\|[0-9]+","",colnames(exprSet)[2:9])

## survial

library(survival)

library(survminer)

my.surv <- Surv(exprSet$times, exprSet$patient.vital_status)## fist get the surv object

log_rank_p <- apply(exprSet[,2:9], 2, function(value1){

  group <- ifelse(value1 > median(value1),"high","low")

  kmfit <- survfit(my.surv ~ group,data = exprSet)

  data.survdiff <- survdiff(my.surv ~ group)

  p.value <- 1 - pchisq(data.survdiff$chisq, length(data.survdiff$n) - 1)

})

log_rank_p < 0.05

names(exprSet)

### select one of the genes in the genelist and plot survival curve

names(exprSet[2:9])

gene = "NOTCH1"

plot_gene <- function(genename = gene){

  group <- sapply(exprSet[,match(genename,names(exprSet))][1],function(x){ifelse(x > median(exprSet[[genename]]),"high","low")})

  exprSet$group <<- group

  kmfit <- survfit(my.surv ~group,data = exprSet)

  invisible(kmfit)

}

plot_gene() %>% ggsurvplot(conf.int=FALSE, pval=TRUE )  + ggtitle(gene)

#### save files

tmp <- log_rank_p %>% data.frame(pvalue = .)

write.csv(tmp,"survival_p_value.csv",quote = T,sep = ",")