small.cell.project: small.cell.project

#' create tcga boxplots
#'
#' Creates boxplots from prediction scores from pls or plsda
#' Makes some boxplots
#'
#' @param dir directory where PLSR predictions are
#' @param pattern pattern to grep for in files that have predictions
#' @param train_string string that has information on the training set, make sure to describe the # of featres
#' @param output_folder the folder to output to, default is ./ i.e. current folder
#' 
#' @importFrom mixOmics pls 
#' @importFrom data.table rbindlist
#' @export
#'

create_tcga_boxplots=function(dir,pattern,output_string,output_folder="./"){
  col_vector=c("darkseagreen4","indianred1","violetred","tomato2","darkcyan","springgreen3","steelblue1","skyblue","brown","royalblue3","rosybrown3",
               "red3","purple2","darkblue","limegreen","deepskyblue","darkviolet","purple4","burlywood1","magenta1","maroon3","indianred3",
               "cornflowerblue","darkorange1","darkseagreen2","dodgerblue1","firebrick2","cadetblue3","purple2","blue1","chocolate",
               "darkslategray1","aquamarine","palevioletred","chartreuse2","forestgreen","orange3","darkolivegreen2","mediumpurple4","gold")
  all.files.short.path=list.files(dir,pattern=pattern,full.names=F)
  all.files.long.path=list.files(dir,pattern=pattern,full.names=T)
  all.names=as.character(sapply(all.files.short.path, function(x) strsplit(x,"_")[[1]][1]))
  list_of_dataframes<-lapply(all.files.long.path, read.table,header=T,row.names=NULL,sep = '\t',check.names=F,stringsAsFactors = F)
  names(list_of_dataframes)=all.names
  for(i in 1:length(all.files.short.path)) {
    list_of_dataframes[[i]]$type=all.names[i]
  }
  list_of_dataframes_scaled=lapply(list_of_dataframes,function (x) {
    cols=ncol(x)
    blah=cbind (x[,1],scale(x[,2:(cols-1)]),x[,cols,drop=F])
    colnames(blah)[[1]]="Sample"
    blah
  })
  output<-data.frame(data.table::rbindlist(list_of_dataframes))
  output.scaled<-data.frame(data.table::rbindlist(list_of_dataframes_scaled))
  
  
  output.scaled.by.top3.avg=output.scaled %>% dplyr::group_by(type) %>% dplyr::arrange(type,dplyr::desc(comp.1)) %>% dplyr::slice(1:3) %>% dplyr::summarise(avg=mean(comp.1)) %>% dplyr::arrange(dplyr::desc(avg))
  scaled.by.top.3= unique(output.scaled.by.top3.avg$type)
  scaled.bymean=levels(with(output.scaled, reorder(type, -comp.1, mean)))
  scaled.bymax=rev(levels(with(output.scaled, reorder(type, comp.1, max))))
  
  unscaled.bymean=levels(with(output, reorder(type, -comp.1, mean)))
  normalize01 <- function(x){
    return((x-min(x)) / (max(x)-min(x)))
  }
  ## TCGA data unscaled 
  set.seed(79)
  unscaled.pic<-ggplot(data=output,aes(x = factor(type,levels=unscaled.bymean), y = comp.1))+geom_boxplot(outlier.color=NA)+
    geom_jitter(width=0.1,aes(color=factor(type)))+theme_bw()+
    theme(axis.text=element_text(size=10,face="bold",angle=90),axis.title=element_text(size=8,face="bold"),legend.position="none",panel.grid.major = element_blank(),
          panel.grid.minor = element_blank(),
          panel.border = element_blank(),
          panel.background = element_blank(), axis.line = element_line(colour = "black", size = 1, linetype = "solid"))+
    scale_color_manual(values =  sample(col_vector,size=40))
  ggsave(paste0(output_folder,output_string,"_TCGA_unscaled_by_mean.png"),
         dpi = 300, plot = unscaled.pic, width = 8, height = 5)
  #TCGA scaled by top 3
  set.seed(79)

  
  axis.line = element_line(colour = "darkblue", 
                           size = 1, linetype = "solid")
  
  scaled.pic<-ggplot(data=output.scaled,aes(x = factor(type,levels=scaled.by.top.3), y = comp.1))+
     geom_boxplot(outlier.color=NA)+
     geom_jitter(width=0.1,aes(color=factor(type)))+ #theme_bw()+
     theme(axis.text=element_text(size=10,face="bold",angle=90),axis.title=element_text(size=8,face="bold"),legend.position="none",
           panel.grid.major = element_blank(),
           panel.grid.minor = element_blank(),
           panel.border = element_blank(),
           panel.background = element_blank(), 
           axis.line = element_line(colour = "black", size = 1, linetype = "solid"))+
    scale_color_manual(values =  sample(col_vector,size=40))


  ggsave(paste0(output_folder,output_string,"_TCGA_scaled_by_avg_top3.png"),
         dpi = 300, plot = scaled.pic, width = 8, height = 5)
}