added in MPNST proteomics

Author: sgosline

build/mpnst/01_mpnst_get_omics.R

@@ -44,11 +44,13 @@ manifest<-synapser::synTableQuery("select * from syn53503360")$asDataFrame()|>
 ##they each get their own sample identifier
 pdx_data<-manifest|>dplyr::select(common_name,starts_with("PDX"))|>
     left_join(pdx_samps)|>
-    dplyr::select(improve_sample_id,RNASeq='PDX_RNASeq',Mutations='PDX_Somatic_Mutations',CopyNumber='PDX_CNV')
+    dplyr::select(improve_sample_id,RNASeq='PDX_RNASeq',Mutations='PDX_Somatic_Mutations',CopyNumber='PDX_CNV',Proteomics='PDX_Proteomics')
 
 tumor_data<- manifest|>dplyr::select(common_name,starts_with("Tumor"))|>
     left_join(tumor_samps)|>
-    dplyr::select(improve_sample_id,RNASeq='Tumor_RNASeq',Mutations='Tumor_Somatic_Mutations',CopyNumber='Tumor_CNV')
+    dplyr::select(improve_sample_id,RNASeq='Tumor_RNASeq',Mutations='Tumor_Somatic_Mutations',CopyNumber='Tumor_CNV')|>
+    mutate(Proteomics='') ##we dont have tumor proteomics from these samples
+#print(tumor_data)
 
 
 combined<-rbind(pdx_data,tumor_data)|>distinct()
@@ -58,6 +60,33 @@ genes_df <- fread(genefile)
 
 
 
+
+##added proteomics first
+proteomics<-do.call('rbind',lapply(setdiff(combined$Proteomics,c('',NA)),function(x){
+                                        # if(x!=""){
+    #print(x)
+    sample<-subset(combined,Proteomics==x)
+    #print(sample)
+    res<-fread(synGet(x)$path)|>
+        #tidyr::separate(Name,into=c('other_id','vers'),sep='\\.')|>
+                                        #dplyr::select(-vers)|>
+        dplyr::rename(gene_symbol='Gene')|>
+        left_join(genes_df)|>
+        dplyr::select(entrez_id,proteomics='logRatio')|>
+        distinct()|>
+        subset(!is.na(entrez_id))|>
+        subset(proteomics!=0)
+
+    res$improve_sample_id=rep(sample$improve_sample_id[1],nrow(res))
+    res$source=rep('NF Data Portal',nrow(res))
+    res$study=rep('MPNST PDX MT',nrow(res))
+    return(distinct(res))
+                                        # }
+}))
+
+fwrite(proteomics,'/tmp/mpnst_proteomics.csv.gz')
+
+
 #### FIRST WE GET RNASeq Data
 
 rnaseq<-do.call('rbind',lapply(setdiff(combined$RNASeq,NA),function(x){
@@ -146,4 +175,3 @@ cnv<-do.call(rbind,lapply(setdiff(combined$CopyNumber,NA),function(x){
 fwrite(cnv,'/tmp/mpnst_copy_number.csv.gz')
 
 ##TODO: get proteomics!!!
-