In sgosline/mpnstXenoModeling: MPNST Xenograft Modeling

knitr::opts_chunk$set(echo = TRUE)
require(remotes)
if(!require(mpnstXenoModeling)){
  remotes::install_github('sgosline/mpnstXenoModeling')
  library(mpnstXenoModeling)
}
library(dplyr)

Here we have an example of using the Xeva package to search for any molecular indicates of drug response in the MPNST PDX data.

Load PDX Data

PDX data is available on this Synapse project page that collates available PDX and tumor model data across the project.

loadPDXData()

Mutational data

Now we can look at the mutational data in more detail

 mutMat<-varData%>%
  mutate(AD=as.numeric(AD))%>%
  subset(Symbol!="")%>%
  dplyr::select(-c(individualID,specimenID,Sample,Age,Sex,MicroTissueQuality,Location,Size,Clinical.Status))%>%
  distinct()%>%
  tidyr::pivot_wider(names_from=synid,values_from=AD,values_fn=list(AD=mean),
                     values_fill=list(AD=0.0))%>%
    tibble::column_to_rownames('Symbol')

library(pheatmap)
  annotes<-varData%>%
    dplyr::select(c(synid,Sample,Age,Sex,MicroTissueQuality,Location,Size,Clinical.Status))%>%
    mutate(MicroTissueQuality=unlist(MicroTissueQuality))%>%
    mutate(`Clinical Status`=unlist(Clinical.Status))%>%
    dplyr::select(-Clinical.Status)%>%
    #tidyr::separate(specimenID,into=c('patient','sample'),remove=TRUE)%>%
    distinct()%>%
    tibble::column_to_rownames('synid')#%>%
#    dplyr::filter(sample!='normal')

  mutMat<-mutMat[,rownames(annotes)]
 pheatmap(log10(0.01+mutMat),clustering_distance_cols = 'correlation',cellwidth = 10,annotation_col = annotes,labels_row = rep("",nrow(mutMat)),labels_col=rep("",ncol(mutMat)))

 pheatmap(log10(0.01+mutMat),clustering_distance_cols = 'correlation',cellwidth = 10,annotation_col = annotes,labels_row = rep("",nrow(mutMat)),labels_col=rep("",ncol(mutMat)),filename='allMutations.pdf')

Top mutations

We then select for those mutations that exist in more than one sample. And then add back TP53.

topMuts=dplyr::filter(varData,AD>0)%>%
    group_by(Symbol)%>%
    summarize(nSamps=n_distinct(individualID))%>%
    dplyr::filter(nSamps>1)%>%
    dplyr::select(Symbol)

topMuts<-intersect(rownames(mutMat),union("TP53",topMuts$Symbol))

 pheatmap(log10(0.01+mutMat[topMuts,]),cellwidth = 10,cellheight=10,annotation_col = annotes,labels_col=rep("",ncol(mutMat)),clustering_method='ward.D2', clustering_distance_cols = 'correlation')
 pheatmap(log10(0.01+mutMat[topMuts,]),cellwidth = 10,cellheight=10,annotation_col = annotes,filename='recMutations.pdf',clustering_method='ward.D2', clustering_distance_cols = 'correlation')

Selected mutations

Below is a list of custom genes we want to evaluate.

genelist=c('NF1','CDKN2A','TP53','EED','SUZ12','EGFR','PDGFRA','MET','BRAF',
           'TYK2','PIK3CA','AURKA2','ATRX','TSC1','TSC2','NF2')
missed.genes<-setdiff(genelist,rownames(mutMat))
print(paste('Missing mutations in genes',paste(missed.genes,collapse=',')))

missed.mat<-matrix(0,nrow=length(missed.genes),ncol=ncol(mutMat),
                   dimnames=list(missed.genes,colnames(mutMat)))

new.mat<-rbind(mutMat[intersect(genelist,rownames(mutMat)),],missed.mat)

 pheatmap(log10(0.01+new.mat),cellwidth = 10,cellheight=10,annotation_col = annotes,labels_col=rep("",ncol(mutMat)),clustering_method='ward.D2')
 pheatmap(log10(0.01+new.mat),cellwidth = 10,cellheight=10,annotation_col = annotes,filename='selectedMutations.pdf',clustering_method='ward.D2')