#R code below

library(data.table)
library(dplyr)
library(pbapply)
library(base)

allexons <- fread("/Users/catherinez/Research/JMPSummit/teich_T_cells_psi_values_labels_final_clean.csv")
top10 <- fread("/Users/catherinez/Research/JMPSummit/bootstrap_forest_top10exons_40trees.csv")

creatematrix <- function(df) {
  df1 <- df[,-1]
  dfm <- matrix(unlist(df1), ncol = ncol(df1), nrow = nrow(df1))
  names(df)[1] <- "id"
  rownames(dfm) <- df$id
  colnames(dfm) <- colnames(df[,-1])
  return(dfm)
}

allexons.m <- creatematrix(allexons)
allexons.t <- t(allexons.m)
allexons.df <- as.data.frame(allexons.t)
allexons.df <- tibble::rownames_to_column(allexons.df, "exons")

top10clean <- top10[,-c(2:4)]
names(top10clean)[1] <- "exons"

top10.df <- merge(x=top10clean,y=allexons.df,by='exons',all.x=TRUE)
top10.final.m <- creatematrix(top10.df)
top10.final.t <- t(top10.final.m)

write.csv(top10.final.t, "/Users/catherinez/Research/JMPSummit/teich_T_cells_top10_bootstrap_forest.csv")