R/GEX_phenotype_per_clone.R

In Platypus: Single-Cell Immune Repertoire and Gene Expression Analysis

#'Plotting of GEX phenotype by VDJ clone
#'
#'@description Integrates VDJ and gene expression libraries by providing cluster membership seq_per_vdj object and the index of the cell in the Seurat RNA-seq object.
#' ! For platypus.version == "v3" and VDJ_GEX_matrix output the function will iterate over entries in the sample_id column of the GEX by default.
#' @param GEX For platypus.version == "v3" the GEX object from the output of the VDJ_GEX_matrix function (VDJ_GEX_matrix.output \[\[2\]\]). For platypus.version == "v2" a single seurat object from automate_GEX function after labeling cell phenotypes using the GEX_phenotype function.
#' @param clonotype.ids  For platypus.version == "v2" Output from either VDJ_analyze or VDJ_clonotype functions. This list should correspond to a single GEX.list object, in which each list element in clonotype.list is found in the GEX.object. Furthermore, these repertoires should be found in the automate_GEX library.
#' @param GEX.group.by For platypus.version == "v3". Character. Column name of the GEX@meta.data to group barplot by. Defaults to seurat_clusters
#' @param GEX.clonotypes For platypus.version == "v3". Numeric vector with ids of clonotypes to plot e.g. c(1,2,3,4). Can also be set to "topclones"
#' @param global.clonotypes Boolean. Defaults to FALSE. Set to True if clonotyping has been done across samples
#' @param platypus.version Set to either "v2" or "v3" depending on wether suppyling GEX_automate or VDJ_GEX_matrix\[\[2\]\] objects. Defaults to "v3"
#' @return Returns a stacked barplot that visualizes the seurat cluster membership for different cell phenotypes.
#' @export
#' @examples
#' small_vgm_cl <- Platypus::small_vgm
#' small_vgm_cl[[2]]$clonotype_id_10x <- "clonotype1"
#' GEX_phenotype_per_clone(GEX = small_vgm_cl[[2]]
#' , GEX.clonotypes = c(1), GEX.group.by = "seurat_clusters", platypus.version = "v3")
#'

GEX_phenotype_per_clone <- function(GEX,
                                    clonotype.ids,
                                    global.clonotypes,
                                    GEX.group.by,
                                    GEX.clonotypes,
                                    platypus.version){

  value <- NULL
  variable <- NULL
  clonotype_id_10x <- NULL
  clonotype_id <- NULL

  if(!missing(clonotype.ids)) platypus.version <- "v3"

  if(!missing(GEX.clonotypes)) platypus.version <- "v3"
  if(missing(global.clonotypes)) global.clonotypes <- FALSE
  if(platypus.version == "v2"){

    seurat.object <- GEX

  # dca<-dcast(seurat.object@meta.data,clonotype_id~cell.state,value.var="cell.state",length)
  # dca$Sum <- rowSums(dca[-1])
  # dca[2:(ncol(dca)-1)]<- dca[2:(ncol(dca)-1)]/dca$Sum
  # dca<-melt(dca[-ncol(dca)], id.vars = "clonotype_id")
  # dca<-dca[which(dca$clonotype_id %in% clonotype.ids),]
  desired_strings <- paste0("clonotype",clonotype.ids)
  possible_states <- unique(seurat.object$cell.state)
  temp.matrix <- as.data.frame(matrix(0,nrow=length(clonotype.ids),ncol=length(possible_states)))
  for(i in 1:nrow(temp.matrix)){
    for(j in 1:ncol(temp.matrix)){
      temp.matrix[i,j] <- length(which(seurat.object$cell.state[which(seurat.object$clonotype_id==desired_strings[i])]==possible_states[j]))/length(seurat.object$cell.state[which(seurat.object$clonotype_id==desired_strings[i])])
    }
  }
  rownames(temp.matrix) <- desired_strings
  colnames(temp.matrix) <- possible_states
  temp.matrix$clonotype_id <- desired_strings

  temp.melt <- reshape2::melt(temp.matrix,id.vars = "clonotype_id")
  stacked.ggplot<-ggplot2::ggplot(data=temp.melt, ggplot2::aes(x=clonotype_id, y=value, fill=variable)) +
    ggplot2::geom_bar(stat="identity")+
    ggplot2::ylab("Cell Counts")+
    ggplot2::xlab("Clonotypes")+
    ggplot2::labs(fill = "Cell State")
  stacked.ggplot <- stacked.ggplot + cowplot::theme_cowplot()+ ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5)) + ggplot2::scale_y_continuous(expand = c(0,0)) + ggplot2::ylab("Number of cells") + ggplot2::xlab("Clonal rank")
  return(stacked.ggplot)

  } else if(platypus.version == "v3"){

    if(global.clonotypes){GEX@meta.data$sample_id <- 1}

    plot.out.list <- list()
    for(k in 1:length(unique(GEX@meta.data$sample_id))){

    GEX@meta.data$iscursample <- NULL
    GEX@meta.data$iscursample <- GEX@meta.data$sample_id == unique(GEX@meta.data$sample_id)[k]

    seurat.object <- subset(GEX, cells = colnames(GEX)[which(GEX$iscursample ==TRUE)])

    #get clonotypes to plot
    if(GEX.clonotypes[[1]] == "topclones"){ #choose 10 top expanded clones
      temp_choice <- seurat.object@meta.data[,c("clonotype_id_10x", "clonotype_frequency")]
      temp_choice <- subset(temp_choice, is.na(clonotype_id_10x) == FALSE)
      temp_choice$clonotype_frequency <- as.numeric(temp_choice$clonotype_frequency)
      temp_choice <- temp_choice[order(temp_choice$clonotype_frequency, decreasing = TRUE),]

      temp_choice <- temp_choice[c(1:10),]
      desired_strings <- unique(temp_choice$clonotype_id_10x)
    } else { desired_strings <- paste0("clonotype",GEX.clonotypes)}
    if(any(!desired_strings %in% seurat.object$clonotype_id_10x)) stop("At least one provided clonotype does not exist in the GEX object")
    #get group ides to plot
    seurat.object$cell.state <- seurat.object@meta.data[,GEX.group.by]

    if(any(is.na(seurat.object$cell.state))){
    #subset further to exclude NA values in the grouping column
    seurat.object$groupisna <- is.na(seurat.object$cell.state) ==TRUE
    seurat.object <- subset(seurat.object, cells = colnames(seurat.object)[which(seurat.object$groupisna ==TRUE)])
    }

    possible_states <- unique(seurat.object$cell.state)
    temp.matrix <- as.data.frame(matrix(0,nrow=length(desired_strings),ncol=length(possible_states)))
    for(i in 1:nrow(temp.matrix)){
      for(j in 1:ncol(temp.matrix)){
        temp.matrix[i,j] <- length(which(seurat.object$cell.state == possible_states[j] & seurat.object$clonotype_id_10x==desired_strings[i]))/length(seurat.object$cell.state[which(seurat.object$clonotype_id_10x==desired_strings[i])]) *100
      }
    }

    rownames(temp.matrix) <- desired_strings
    colnames(temp.matrix) <- possible_states
    temp.matrix$clonotype_id <- desired_strings


    temp.melt <- reshape2::melt(temp.matrix,id.vars = "clonotype_id")

    if(is.null(levels(seurat.object$cell.state)) == FALSE){
    temp.melt$variable <- ordered(as.factor(temp.melt$variable), levels = levels(seurat.object$cell.state))
    }

    temp.melt$clonotype_id <- ordered(as.factor(temp.melt$clonotype_id), levels = desired_strings)

    stacked.ggplot<- ggplot2::ggplot(data=temp.melt, ggplot2::aes(x=clonotype_id, y=value, fill=variable)) +
      ggplot2::geom_bar(stat="identity")+
      ggplot2::ylab("Cell Counts")+
      ggplot2::xlab("Clonotypes")+
      ggplot2::labs(fill = "Cell State")
    stacked.ggplot <- stacked.ggplot + cowplot::theme_cowplot() + ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5)) + ggplot2::scale_y_continuous(expand = c(0,0)) + ggplot2::ylab("Number of cells") + ggplot2::xlab("Clonal rank") +ggplot2::scale_fill_manual(values = grDevices::rainbow(length(unique(temp.melt$variable)))) + ggplot2::ggtitle(label = paste0(unique(GEX@meta.data$sample_id)[k]))
    plot.out.list[[k]] <- stacked.ggplot
    } #end loop over sample_ids
    return(plot.out.list)
  }
}