#load libs
library( knitr )
library(ANTsR)
library(visreg)
library(robustbase)
library(groupdata2)
library(ggplot2)
library(caret)
require(broom)
#Manganese Enhanced MRI predicts cognitive performnace Alex Badea and Natalie Delpratt
#reuses results of sparsedecom2 for best performing fold Alex Badea 7 dept 2018
#uses RMSE as prediction error, rather than goodness of fit error 30 August 2018
#legacy from Natalie to rememebr paths
#setwd( '/Users/omega/alex/adforesight/' )
#output.path <- '/Users/omega/alex/adforesight/mydata/outdata/sd2_projall_noscale/' #ND
mypath<-'/Volumes/CivmUsers/omega/alex/GitHub/adforesight/'
mypath <- '/Users/alex/GitHub/adforesight/' #flavors of serifos
mypath <- '/Users/alex/Documents/GitHub/adforesight/' #flavors of ithaka
setwd(mypath)
source(paste(mypath, '/R/myr2score.R',sep=''))
mysp <- 0.05 #0.05 # 0.01 # 0.05 #0.2 #0.05 # was 0.005 sparseness
#if mynvecs is set to one eig1 neetds to be transposed
mynvecs <- 2 # 5 vecs is better # 10 shows ventral thalamic nuclei 10 # 50 #put back to 50 alex #tested with 10
myell1 <- 1 # make this smaller 0.5, Brian says it is not what i think just switch between l0 and l1
myits<-15 #15 #5
mysmooth<-1 #0.1 # 0.1 #0.01 #0 # was 0.01
myclus<-250 #was 250
# Load in Behavior and Imaging Data
behavior <- read.csv('./mydata/All_Behavior.csv')
labled.set <-read.csv('./mydata/legendsCHASS2symmetric.csv')
labeled.brain.img <- antsImageRead('./mydata/MDT_labels_chass_symmetric.nii.gz')
mask <- antsImageRead('./mydata/MDT_mask_e3.nii')
mask <- thresholdImage( mask, 0.1, Inf )
#read all 3 contrast files
mang_files <- list.files(path = "./mydata/imdata/", pattern = "T2_to_MDT",full.names = T,recursive = T)
jac_files <- list.files(path = "./mydata/imdata/", pattern = "jac_to_MDT",full.names = T,recursive = T)
chi_files <- list.files(path = "./mydata/imdata/", pattern = "X_to_MDT",full.names = T,recursive = T)
########################################
#build a place to save results
extension<-paste('sd2SINGLEChi', 'sp', toString(mysp), 'vecs', toString(mynvecs), 's', toString(mysmooth),'clus', toString(myclus), sep='') # 'JACsp0p005s0'
output.path <- paste(mypath,'/mydata/outdata_sd2/',extension, '/', sep='') #sd2_projall_noscale/'
if (dir.exists(output.path)){ 1} else {dir.create(output.path, recursive=TRUE)}
#pick yourcontrast
mang_mat <- imagesToMatrix(chi_files,mask)
#######################################
#let things flow from here
mygroup <- behavior$genotype[1:24]
myindex <- c(1:24)
mydfb <- data.frame("mysubject_index" = factor(as.integer(myindex)),"mygenotype"=mygroup)
kable(mydfb, align = 'c')
set.seed(1)
k<-4
performances <- c()
myBICs <- c()
myR2score<-c()
myps<-c()
gfit<-c()
###build k models and retain the best performing one in terms of RMSE2
#considet using LOOCV to replace folds, but results may be unstable
#k<-length(rows.train)-1
k<-4
set.seed(1)
res_train<-createFolds(behavior$genotype,k, list = TRUE, returnTrain = TRUE)
set.seed(1)
res_test<-createFolds(behavior$genotype,k)
for (myfold in 1:k){
# for (myfold in 3){
gc(verbose = TRUE, reset = FALSE)
print('myfold:',myfold)
print(myfold)
rows.train<-as.integer(unlist(res_train[myfold]))
rows.test<-as.integer(unlist(res_test[myfold]))
mang.train <- mang_mat[rows.train, ]
mang.test <- mang_mat[rows.test, ]
behav.train <- behavior[rows.train, ]
behav.test <- behavior[rows.test, ]
dist4.train <- behav.train[,'d4']
dist4.test <- behav.test[,'d4']
start_time <- Sys.time()
#negative sparseness is what? allows for negative weights!
myeig2_mang<-sparseDecom2(inmatrix = list(mang.train,as.matrix(behav.train$d4)),its = myits, cthresh=c(myclus,0), smooth = mysmooth, mycoption = 0, sparseness = c(mysp,1), nvecs = mynvecs, verbose=1, statdir=paste(output.path))
#myeig2_mang<-sparseDecom(inmatrix = mang.train,its = myits, cthresh=c(myclus), smooth = mysmooth, mycoption = 0, sparseness = c(mysp), nvecs = mynvecs, verbose=1, statdir=paste(output.path2))
end_time <- Sys.time()
t1time<-end_time - start_time
print(t1time)
imgpredtrain_mang<-mang.train %*% (myeig2_mang$eig1)
imgpredtest_mang<-mang.test %*% (myeig2_mang$eig1)
####start do single alex
ncolcombo<-ncol( imgpredtrain_mang)
projs.train <- data.frame(dist4.train, imgpredtrain_mang) # column combind the behavior wth the projections
colnames(projs.train) <- c('Dist_4', paste0('Proj', c(1:ncolcombo)))
projs.test <- data.frame(dist4.test, imgpredtest_mang ) # column combind the behavior wth the projections
colnames(projs.test) <- c('Dist_4', paste0('Proj', c(1:ncolcombo)))
###end do single alex
mylm <- lm('Dist_4 ~ .', data=projs.train) # behavior correlation with projections
summylm<-summary(mylm)
summanovalm<-anova(mylm)
rSquared <- summary(mylm)$r.squared
pVal <- anova(mylm)$'Pr(>F)'[1]
mylmsummary<-glance(mylm)
pval1<-mylmsummary$p.value
e2i_mang<-matrixToImages((t(myeig2_mang$eig1)),mask = mask)
for (i in 1:mynvecs){
antsImageWrite(e2i_mang[[i]],paste(output.path,extension,'sd2eig_' ,as.character(i), 'fold', toString(myfold), '_Mn.nii.gz',sep=''))
# antsImageWrite(e2i_jac[[i]],paste(output.path,extension,'sd2eig_' ,as.character(i), 'fold', toString(myfold), '_jac.nii.gz',sep=''))
# antsImageWrite(e2i_chi[[i]],paste(output.path,extension,'sd2eig_' ,as.character(i), 'fold', toString(myfold), '_chi.nii.gz',sep=''))
}
distpred4 <- predict.lm(mylm, newdata=projs.test) # based on the linear model predict the distances for the same day
glance(cor.test(projs.test$Dist_4,(distpred4)))
glance(cor.test(distpred4,dist4.test))
#remove next lines for LOOCV
mymodel<-lm(distpred4~dist4.test)
modsum <-summary(mymodel)
r2 <- modsum$r.squared #modsum$adj.r.squared
# my.p <- modsum$coefficients[2,4]
RMSE2<-sqrt(mean((distpred4 - dist4.test)^2))
performances[myfold]<-RMSE2
myR2score[myfold]<-myr2score(distpred4,dist4.test)
myps[myfold]<-pval1<-mylmsummary$p.value #my.p
myBICs[myfold] <- BIC(mylm)
###
mytheme <- theme(panel.grid.minor = element_blank(), panel.grid.major = element_blank(),
panel.background = element_rect(fill = "white"))
myplot<- visreg(mymodel, gg=TRUE)
myplot + theme(panel.grid.minor = element_blank(), panel.grid.major = element_blank(),
panel.background = element_rect(fill = "transparent", colour = NA),
#xaxs="i", yaxs="i",
axis.line.x = element_line(colour = 'black', size=0.5, linetype='solid'),
axis.line.y = element_line(colour = 'black', size=0.5, linetype='solid')) +
ggtitle(paste("RMSE=",formatC(RMSE2,digits=2, format="f"), "p=",formatC(myps[myfold],digits=4, format="f"), " BIC=", formatC(BIC(mymodel),digits=2, format="f")))
ggsave(paste(output.path,extension,'Mnfold',toString(myfold),'.pdf',sep=''), plot = last_plot(), device = 'pdf',
scale = 1, width = 4, height = 4, units = c("in"),dpi = 300)
save(mylm, file=paste(output.path , "model2", toString(myfold), ".Rdata", sep=''))
save(mymodel, file=paste(output.path , "behavmodelsd2", toString(myfold), ".Rdata", sep=''))
myperf<-data.frame(rbind(distpred4,dist4.test),row.names=c("d_predicted","d_valid"))
write.csv(myperf, file = paste(output.path ,extension,'distances4_pv_fold' , toString(myfold), '.csv',sep=''))
myperf<-data.frame(c(RMSE2,myR2score[myfold],myps[myfold],myBICs[myfold], r2),row.names=c("RMSE2","R2score","p","BIC", "R2"))
write.csv(myperf, file = paste(output.path ,extension,'distances4_stats_fold' , toString(myfold), '.csv',sep=''))
#plot(dist4.test, distpred4, xlab = 'Real Swim Distance on Day 4', ylab = 'Predicted Swim Distance on day 4',
# main='Predicted vs. Real Swim Distance on Day 4', ylim = c(0,1200),xlim = c(0,1200)) # generate plot
#
# for ( bestpred in 1:ncol(imgpredtrain_mang)) {
# cogpredtrain_mang <-behav.train$d4 %*% t(as.matrix(myeig2$eig2)[,bestpred])
# cogpredtest_mang <- behav.test$d4 %*% t(as.matrix(myeig2$eig2)[,bestpred])
# gvars<-paste("Proj",c(1:nrow(myeig2$eig2)),sep='',collapse='+')
# projs.train <- data.frame(cbind(cogpredtrain_mang,imgpredtrain_mang))
# colnames(projs.train) <- c('cognitive',paste0( 'Proj', c( 1:ncol(imgpredtrain_mang ) )))
# projs.test <- data.frame(cbind(cogpredtest_mang, imgpredtest_mang))
# colnames(projs.test) <- c('cognitive',paste0( 'Proj', c( 1:ncol(imgpredtest_mang ) )))
# myform<-as.formula( paste("cognitive~",gvars,sep='') )
# mylm <- lm(myform, data=projs.train)
# distpred <- bigLMStats( mylm)
# cat(paste("Eig",bestpred,"is related to:\n"))
# #mycog<-colnames(behav.train.mat)[ abs(myeig2$eig2[,bestpred]) > 0 ]
# mycog<-colnames(behav.train)[5]
# cat( mycog )
# cat("\nwith weights\n")
# cat( abs(myeig2$eig2[,bestpred])[ abs(myeig2$eig2[,bestpred]) > 0 ])
# cat(paste("\nwith predictive correlation:", cor( cogpredtest_mang,predict(mylm,newdata=projs.test))))
# }
#gc(verbose = TRUE, reset = FALSE, full = TRUE)
gc(verbose = TRUE, reset = FALSE)
}
###################################
##### for validation now ####
###################################
myminfold<-which(performances == min(performances), arr.ind = TRUE)
myfold<-myminfold
load(file=paste(output.path , "model2", toString(myfold), ".Rdata", sep='')) # loads mylm
ncolcombo<-mynvecs
rows.valid <- c(1:24)
mang.valid <- mang_mat[rows.valid, ]
#read eigenregions for best myfold
#paste(output.path,extension,'sd2eig' ,as.character(i), 'fold', toString(myfold), '.nii.gz',sep='')
eig_files_Mn <- list.files(path = paste(output.path,sep=''), pattern=paste('*', 'fold', toString(myfold), '_Mn.nii.gz', sep=''),full.names = T,recursive = T)
eig_mat_Mn <- imagesToMatrix(eig_files_Mn,mask)
imgmat_mang_valid <- mang.valid %*% t(eig_mat_Mn) # [24,numvox] [nvecsx3,numvox]
dist4.valid <- behavior[rows.valid, 'd4']
projs.valid <- data.frame(cbind(dist4.valid,imgmat_mang_valid))
colnames(projs.valid) <- c('Dist_4', paste0('Proj', c(1:ncolcombo)))
distpred <- predict.lm(mylm, newdata=projs.valid)
mymodel<-lm(distpred~dist4.valid)
RSS <- c(crossprod(mymodel$residuals))
MSE <- RSS / length(mymodel$residuals)
RMSE <- sqrt(MSE)
RMSE2<-sqrt(mean((distpred - dist4.valid)^2))
mysummary <-summary(mymodel)
r2pred <- mysummary$adj.r.squared
ppred <- mysummary$coefficients[2,4]
max(behavior$d4[1:24])
RMSE_valid<-RMSE2
BIC_valid <- BIC(mymodel)
R2score_valid<-myr2score(distpred,dist4.valid)
res_cor<-cor.test(dist4.valid,distpred)
myplot<- visreg(mymodel, gg=TRUE, scale='linear', plot=TRUE, xlim=c(0,max(dist4.valid)),ylim=c(0,max(dist4.valid)))
myplot2<-plot(myplot,xlim=c(0,max(dist4.valid)),ylim=c(0,max(dist4.valid)))
ggsave(paste(output.path,extension,'MnValidationSet',toString(myfold),'sd2plainjane.pdf',sep=''), plot = last_plot(), device = 'pdf',
scale = 1, width = 4, height = 4, units = c("in"),dpi = 300)
myplot + theme(panel.grid.minor = element_blank(), panel.grid.major = element_blank(),
panel.background = element_rect(fill = "transparent", colour = NA),
axis.line.x = element_line(colour = 'black', size=0.5, linetype='solid'),
axis.line.y = element_line(colour = 'black', size=0.5, linetype='solid'))+
#xlim(0,1200)+ylim(0,1200)+coord_cartesian(xlim = c(1200, 1200),ylim = c(1200, 1200)) + coord_equal()+
ggtitle(paste("RMSE=",formatC(RMSE_valid,digits=2, format="f"),
# "R2score=",formatC(R2score_valid,digits=2, format="f"),
# " R2=", formatC(r2pred,digits=2, format="f"),
" p= ", formatC(ppred,digits=4, format="f"),
" BIC=", formatC(BIC_valid,digits=2, format="f")))
ggsave(paste(output.path,extension,'MnValidationSet',toString(myfold),'sd2.pdf',sep=''), plot = last_plot(), device = 'pdf',
scale = 1, width = 4, height = 4, units = c("in"),dpi = 300)
numcols<-dim(projs.valid)[2]
rd4 <- t(t(dist4.valid)[rep(1,c(3*mynvecs)),])
pcor<-c(numcols)
corval<-c(numcols)
for (i in 1:numcols) {
mypcor<-cor.test(t(dist4.valid),t(projs.valid[,i]))
pcor[i]<-mypcor$p.value
corval[i]<-mypcor$estimate
}
rt<-glance(cor.test(dist4.valid,distpred))
corval[1]<-rt$estimate
pcor[1]<-rt$p.value
# for (i in 1:numcols){
# res1eig<-matrixToImages(jeanat_mang$fusedlist,mask = mask)[[i]] #eig1
# antsImageWrite(res1eig,paste(output.path,extension,'JoinEanat' ,as.character(i), '.nii.gz',sep=''))
# }
mycorsdf_eig2d4<-data.frame(rbind(pcor,corval),row.names=c("pcor","cor"))
colnames(mycorsdf_eig2d4)<-c('total', paste0('Proj', c(1:ncolcombo)))
write.csv(mycorsdf_eig2d4, file = paste(output.path ,extension,'fold', toString(myfold), 'd4corsvalidsd2.csv',sep=''))
myperf<-data.frame(rbind(distpred,dist4.valid),row.names=c("d_predicted","d_valid"))
write.csv(myperf, file = paste(output.path ,extension,'fold', toString(myfold), 'distances4_validsd2.csv',sep=''))
myeig2_mang_valid<-sparseDecom(inmatrix = mang.valid,its = myits, cthresh=c(myclus), smooth = mysmooth, mycoption = 0, sparseness = c(mysp), nvecs = mynvecs, verbose=1, statdir=paste(output.path2))
e2i_mang_valid<-matrixToImages(((myeig2_mang_valid$eigenanatomyimages)),mask = mask)
for (i in 1:mynvecs){
antsImageWrite(e2i_mang_valid[[i]],paste(output.path,extension,'full_eig' ,as.character(i), '_Mn.nii.gz',sep=''))
}
#redo fold min or save models
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.