###############################################
#   H LEE MOFFITT CANCER CENTER - BIOSTATISTICS CORE - Program
#   PROJECT NAME	 : Bad Path Paper
#   PI			 : Johnaton Lancaster
#   STATISTICIANS	 : Dung-Tsa Chen and William Fulp
#   SOFTWARE/VERSION     : R/Version 2.6.2
#   INFRASTRUCTURE       : PC
###############################################

##Any questions about the code please email William Fulp (william.fulp@moffitt.org)

###Change directory to where datasets were unzipped;

##May need to download Design Library
library(Design);

####Moffitt Duke Ovarian Cancer Dataset;
load(file='Moffitt.Duke.FF.142.RData');

surv.obj <- Surv(Moffitt.Duke.FF.142.Histology.Dataset$Survival.Years,Moffitt.Duke.FF.142.Histology.Dataset$Vital.Status=='Dead')
pca.obj <- prcomp(t(Moffitt.Duke.FF.142.Gene.Dataset),scale.=TRUE)
pca.vals <- pca.obj$x[,1];
med.group <- factor(pca.vals>median(pca.vals),levels=c(F,T),labels=c('Low Score Group','High Score Group'))
pca.surv.out <- survfit(surv.obj~med.group);
pca.surv.p <- pchisq(survdiff(surv.obj~med.group)$chisq,1,lower.tail=F);
pca.cox.p <- summary(coxph(surv.obj~pca.vals))$logtest[3]
survplot(pca.surv.out,time.inc=2,xlab='Years',conf='none',n.risk=T,col=1:2);
title(sub=list(paste('Log-Rank Test p=',round(pca.surv.p,4),'; Cox Model p=',round(pca.cox.p,4),';',sep=''),cex=1))

debulk.med.group <- factor(paste(Moffitt.Duke.FF.142.Histology.Dataset$Debulk,med.group,sep='/'),levels=c('O/Low Score Group','O/High Score Group','S/Low Score Group','S/High Score Group'))
log.rank.p <- pchisq(survdiff(surv.obj~debulk.med.group)$chisq,3,lower.tail=F);
survplot(survfit(surv.obj~debulk.med.group),time.inc=2,xlab='Years',conf='none',n.risk=T,col=1:2);
title(sub=list(paste('Log-Rank Test p=',round(log.rank.p,4),';',sep='')));

cis.med.group <- factor(paste(Moffitt.Duke.FF.142.Histology.Dataset$CIS,med.group,sep='/'),levels=c('IR/Low Score Group','IR/High Score Group','CR/Low Score Group','CR/High Score Group'))
log.rank.p <- pchisq(survdiff(surv.obj~cis.med.group)$chisq,3,lower.tail=F);
survplot(survfit(surv.obj~cis.med.group),time.inc=2,xlab='Years',conf='none',n.risk=T,col=1:2)
title(sub=list(paste('Log-Rank Test p=',round(log.rank.p,4),';',sep='')));

temp.index <- debulk.med.group=='O/Low Score Group' | debulk.med.group=='S/High Score Group'
pchisq(survdiff(surv.obj[temp.index,]~debulk.med.group[temp.index])$chisq,1,lower.tail=F);

temp.index <- debulk.med.group=='O/Low Score Group' | debulk.med.group=='O/High Score Group'
pchisq(survdiff(surv.obj[temp.index,]~debulk.med.group[temp.index])$chisq,1,lower.tail=F);

temp.index <- debulk.med.group=='S/Low Score Group' | debulk.med.group=='S/High Score Group'
pchisq(survdiff(surv.obj[temp.index,]~debulk.med.group[temp.index])$chisq,1,lower.tail=F);

temp.index <- debulk.med.group=='S/Low Score Group' | debulk.med.group=='O/High Score Group'
pchisq(survdiff(surv.obj[temp.index,]~debulk.med.group[temp.index])$chisq,1,lower.tail=F);

coxph(surv.obj~pca.vals+CIS+Age.at.Dx+Debulk+Grade,data=Moffitt.Duke.FF.142.Histology.Dataset);


####GSE9891 Cancer Dataset;
load(file='GSE9891.RData');

surv.obj <- Surv(GSE9891.Histology.Dataset$Survival.Years,GSE9891.Histology.Dataset$Vital.Status=='Dead')

pca.obj <- prcomp(t(GSE9891.Gene.Dataset),scale.=TRUE)
pca.vals <- pca.obj$x[,1];
pca.group <- factor(pca.vals>median(pca.vals),levels=c(F,T),labels=c('Low Score Group','High Score Group'))
pca.surv.out <- survfit(surv.obj~pca.group);
pca.surv.p <- pchisq(survdiff(surv.obj~pca.group)$chisq,1,lower.tail=F);
pca.cox.p <- summary(coxph(surv.obj~pca.vals))$logtest[3]
survplot(pca.surv.out,time.inc=2,xlab='Years',conf='none',n.risk=T,col=1:2);
title(sub=list(paste('Log-Rank Test p=',round(pca.surv.p,4),'; Cox Model p=',round(pca.cox.p,4),';',sep=''),cex=1))


#####Moffitt Colon 205 Dataset;
load(file='Moffitt.Colon205.RData');

surv.obj <- Surv(Moffitt.Colon205.Clinical.Dataset$Survival.Years,Moffitt.Colon205.Clinical.Dataset$Vital.Status=='Dead');
pca.obj <- prcomp(t(Moffitt.Colon205.Gene.Dataset),scale.=TRUE)
pca.vals <- pca.obj$x[,1];
pca.group <- factor(pca.vals>median(pca.vals),levels=c(F,T),labels=c('Low Score Group','High Score Group'))
surv.out <- survfit(surv.obj~pca.group);
surv.p <- pchisq(survdiff(surv.obj~pca.group)$chisq,1,lower.tail=F);
cox.p <- summary(coxph(surv.obj~pca.vals))$logtest[3]
survplot(surv.out,time.inc=1,xlab='Years',conf='none',n.risk=T,col=1:2,label.curves=T);
title(sub=list(paste('Log-Rank Test p=',round(surv.p,4),'; Cox Model p=',round(cox.p,4),';',sep=''),cex=1))


#####Brain Cancer Dataset (50 obs) Nutt et al;
load(file='Nutt.RData');

surv.obj <- Surv(Nutt.Histology.Dataset$Survival.Months,Nutt.Histology.Dataset$Vital.Status=='Dead')
pca.obj <- prcomp(t(Nutt.Gene.Dataset),scale.=TRUE)
pca.vals <- pca.obj$x[,1];
pca.group <- factor(pca.vals>median(pca.vals),levels=c(F,T),labels=c('Low Score Group','High Score Group'))
pca.surv.out <- survfit(surv.obj~pca.group);
pca.surv.p <- pchisq(survdiff(surv.obj~pca.group)$chisq,1,lower.tail=F);
pca.cox.p <- summary(coxph(surv.obj~pca.vals))$logtest[3]
survplot(pca.surv.out,time.inc=6,xlab='Months',conf='none',n.risk=T,col=1:2,label.curves=TRUE);
title(sub=list(paste('Log-Rank Test p=',round(pca.surv.p,4),'; Cox Model p=',round(pca.cox.p,4),';',sep=''),cex=1))


####Brain Cancer Dataset (182 obs);
load(file='GSE13041.RData');

surv.obj <- Surv(GSE13041.Histology.Dataset$Survival.Years,GSE13041.Histology.Dataset$Vital.Status=='Dead');
pca.obj <- prcomp(t(GSE13041.Gene.Dataset),scale.=TRUE)
pca.vals <- pca.obj$x[,1];
pca.group <- factor(pca.vals>median(pca.vals),levels=c(F,T),labels=c('Low Score Group','High Score Group'))
pca.surv.out <- survfit(surv.obj~pca.group);
pca.surv.p <- pchisq(survdiff(surv.obj~pca.group)$chisq,1,lower.tail=F);
pca.cox.p <- summary(coxph(surv.obj~pca.vals))$logtest[3]
survplot(pca.surv.out,time.inc=1,xlab='Years',conf='none',n.risk=T,col=1:2,label.curves=TRUE);
title(sub=list(paste('Log-Rank Test p=',round(pca.surv.p,4),'; Cox Model p=',round(pca.cox.p,4),';',sep=''),cex=1))


##Wang et al breast cancer relapse free survival study;
load(file='Wang.RData')

surv.obj <- Surv(Wang.Histology.Dataset$Relapse.Years,Wang.Histology.Dataset$Relapse.Status=='Relapse')
pca.obj <- prcomp(t(Wang.Gene.Dataset),scale.=TRUE)
pca.vals <- pca.obj$x[,1];
pca.group <- factor(pca.vals>median(pca.vals),levels=c(F,T),labels=c('Low Score Group','High Score Group'))
pca.surv.out <- survfit(surv.obj~pca.group);
pca.surv.p <- pchisq(survdiff(surv.obj~pca.group)$chisq,1,lower.tail=F);
pca.cox.p <- summary(coxph(surv.obj~pca.vals))$logtest[3]
survplot(pca.surv.out,conf='none',n.risk=T,time.inc=2,col=c(1,2),lty=c(1,2),xlab='Years',label.curves=TRUE)
title(sub=paste('Log-Rank Test p=',round(pca.surv.p,4),'; Cox Model p=',round(pca.cox.p,4),';',sep=''));


##Chanrion et al tamoxifen-treated primary breast cancer study;
load(file='Chanrion.RData')

surv.obj <- Surv(Chanrion.Histology.Dataset$Relapse.Years,Chanrion.Histology.Dataset$Relapse.Status=='Relapse');
pca.obj <- prcomp(t(Chanrion.Gene.Dataset),scale.=TRUE)
pca.vals <- pca.obj$x[,1];
pca.group <- factor(pca.vals>median(pca.vals),levels=c(F,T),labels=c('Low Score Group','High Score Group'))
pca.surv.out <- survfit(surv.obj~pca.group);
pca.surv.p <- pchisq(survdiff(surv.obj~pca.group)$chisq,1,lower.tail=F);
pca.cox.p <- summary(coxph(surv.obj~pca.vals))$logtest[3]
survplot(pca.surv.out,time.inc=1,xlab='Years',conf='none',n.risk=T,col=1:2,label.curves=TRUE);
title(sub=list(paste('Log-Rank Test p=',round(pca.surv.p,4),'; Cox Model p=',round(pca.cox.p,4),';',sep=''),cex=1))