Nothing
R/VDJ_clonal_expansion.R
In Platypus: Single-Cell Immune Repertoire and Gene Expression Analysis
Defines functions
VDJ_clonal_expansion
Documented in
VDJ_clonal_expansion
#' Flexible wrapper for clonal expansion barplots by isotype, GEX cluster etc.
#'
#'@description Clonal frequency plot displaying clonal expansion for either T and B cells with Platypus VDJ_build() output. For Platypus v2 plotting of B cell clonotype expansion and isotypes please refer to VDJ_isotypes_per_clone.
#' @param VDJ VDJ dataframe generated using the VDJ_build function.
#' @param celltype Character. Either "Tcells" or "Bcells". If set to Tcells bars will not be colored by default and the parameters treat_incomplete_cells, treat_incomplete_clones, subtypes and species are ignored. The color.by and group.by arguments work identically for both celltypes. If none provided it will detect this param from the celltype column.
#' @param clones numeric value indicating the number of clones to be considered for the clonal expansion plot. Default value is 50. For a standard plot more than 50 is discouraged. When showing only one - possibly rare - isotype via isotypes.to.plot it may be useful to set this number higher (e.g. 100-200)
#' @param subtypes Logical indicating whether to display isotype subtypes or not.
#' @param isotypes.to.plot Character vector. Defaults to "all". This can be set to any number of specific Isotypes, that are to be shown exclusively. For example, to show only clones containing IgG, input "IGHG". If only wanting to check clones with IgA and IgD input c("IGHA","IGHD"). Works equally if subtypes are set to TRUE. Is ignored if color.by is not set to "isotype"
#' @param species Character indicating whether the samples are from "Mouse" or "Human". Default is "Human".
#' @param treat.incomplete.clones Character indicating how to proceed with clonotypes lacking a VDJC (in other words, no cell within the clonotype has a VDJC). "exclude" removes these clonotypes from the analysis. "include" keeps these clonotypes in the analysis. In the plot they will appear has having an unknown isotype.
#' @param treat.incomplete.cells Character indicating how to proceed with cells assigned to a clonotype but missing a VDJC. "proportional" to fill in the VDJ isotype according to the proportions present in of clonotype (in case present proportions are not replicable in the total number of cells e.g. 1/3 in 10 cells, values are rounded to the next full integer and if the new counts exceed the total number of cells, 1 is subtracted from the isotype of highest frequency. If the number is below the number of cell, 1 is added to the isotype with lowest frequency to preserve diversity), "exclude" to exclude them from analysis and rank clonotypes only by the number of cells with a heavy chain. This ranking may deviate from the frequency column in the clonotype table. CAVE: if treat_incomplete_cells is set to "exclude", clonotypes lacking a VDJC entierly will be removed from the analysis. This results in a similar but not identical output as when treat_incomplete_clones is set to true. The two parameters are thereby non-redundant.
#' @param group.by Character. Defaults to "sample_id". Column name of VDJ to split VDJ by. For each unique entry in that column a plot will be generated. Therefore plots can be generated by sample_id, group_id or any other metadata item.To get plots for the whole repertoire set to "none"
#' @param color.by Character. Defaults to "isotype". If set to "isotype" bars are colored by the respective IgH chain or in grey for T cells. This can alternatively be set to any column name of the VDJ. This allows coloring clones by their V_gene usage or by GEX clusters
#' @param variant.plot Logical indicating whether to plot the output showing the variants or not.
#' @param text.size Numeric value indicating the size of the text in the plot. Default is 12.
#' @return Returns a nested list. out[[1]] are plots out[[2]] are raw datatables containing also barcode and CDR3 information
#' @export
#' @examples
#'clonal_out <- VDJ_clonal_expansion(VDJ = Platypus::small_vdj,
#' celltype = "Bcells", clones = 30,subtypes = FALSE, species = "Mouse"
#' ,treat.incomplete.clones = "exclude"
#' ,treat.incomplete.cells = "proportional")
#'clonal_out[[1]] #list of plots
#'clonal_out[[2]] #list of source dataframes
VDJ_clonal_expansion <- function(VDJ,
celltype,
clones,
subtypes,
isotypes.to.plot,
species,
treat.incomplete.clones,
treat.incomplete.cells,
group.by,
color.by,
variant.plot,
text.size){
#Adding compatibility with input naming scheme
VDJ.matrix <- VDJ
VDJ.matrix[is.na(VDJ.matrix)] <- 'unknown'
VDJ <- NULL
ClonalRank <- NULL
Counts <- NULL
sum_counts <- NULL
Isotype <- NULL
Color <- NULL
pasted_variants <- NULL
isotype <- NULL
colors <- NULL
match_ex_crit <- NULL
clonotype_id <- NULL
variant <- NULL
n <- NULL
#add a color scheme similar to Seurat's
ggplotColours <- function(n, h = c(0, 360) + 15){
if ((diff(h) %% 360) < 1) h[2] <- h[2] - 360/n
grDevices::hcl(h = (seq(h[1], h[2], length = n)), c = 100, l = 65)
}
if(missing(clones)) clones <- 50
if(missing(subtypes)) subtypes <- FALSE
if(missing(isotypes.to.plot)) isotypes.to.plot <- "all"
if(missing(species)) species <- "Human"
if(missing(treat.incomplete.cells)) treat.incomplete.cells <- "proportional"
if(missing(treat.incomplete.clones)) treat.incomplete.clones <- "exclude"
if(missing(text.size)) text.size <- 12
if(!treat.incomplete.cells %in% c("exclude", "proportional")){
stop("Please set treat.incomplete.cells to either 'proportional' or 'exclude'. 'Proportional' will assign cells of a clonotype missing a VDJ chain proportionally to the isotypes present in that clone (Default). 'exclude' will remove all cells missing a VDJ chain and thereby also alter clonotype frequencies")
}
if(!treat.incomplete.clones %in% c("exclude", "include")){
stop("Please set treat.incomplete.clones to either 'include' or 'exclude'. 'include' will show also clones which of which no cell has a VDJ chain. 'exclude' will remove such clones (default)")
}
platypus.version <- "v3"
if(missing(celltype)){
if(stringr::str_detect(VDJ.matrix$celltype[1], "B")){celltype = "Bcells"
} else if(stringr::str_detect(VDJ.matrix$celltype[1], "T")){celltype = "Tcells"
} else {stop("No celltype found in celltype column. celltype column must contain either 'B cells' or 'T cells'")}
}
if(missing(group.by)){
group.by <- "sample_id"}
if(missing(variant.plot)){
variant.plot <- FALSE}
if(group.by == "none"){
group.by <- "ungroup"
VDJ.matrix$ungroup <- 1
} else if(!group.by %in% names(VDJ.matrix) & group.by != "sample_id" & group.by == "none"){
stop("Please provide a valid column name of the VDJ.matrix in group.by")
}
if(missing(color.by)){
color.by <- "isotype"}
if(!color.by %in% names(VDJ.matrix) & color.by != "isotype"){
stop("Please provide a valid column name of the VDJ.matrix in color.by, or set to 'isotype' for default (also for Tcells)")
} else if(color.by %in% names(VDJ.matrix) & color.by != "isotype"){
unique_colors <- ggplotColours(n = length(unique(VDJ.matrix[,color.by])))
names(unique_colors) <- unique(VDJ.matrix[,color.by])
}
if(celltype == "Bcells"){ ####START Bcells
#check that dataframe is the input
if(!inherits(VDJ.matrix,"data.frame")){
stop("Please provide a VDJ dataframe. (VDJ_GEX_matrix_out[[1]]")
}
#set up for subsetting by group.by item
VDJ.matrix[,group.by] <- as.character(VDJ.matrix[,group.by])
sample.names <- unique(VDJ.matrix[,group.by])
sample.names[is.na(sample.names)] <- "NONE"
VDJ.matrix[is.na(VDJ.matrix[,group.by]),group.by] <- "NONE"
VDJ.matrix.all <- VDJ.matrix
#Disabled due to the possibility of having low abundance of isotypes of interest. If filtered by isotype, 50 clones may not be enought to get a decent plot.
#if(clones<1 | clones > 50){stop("Number of clones must be an integer value between 1 and 50")}
VDJ_per_clone_output_all <- list()
clones_per_isotype <- list()
clones_per_isotype_all <-list()
output_plot <- list()
variant_df_list <- list()
if(subtypes == FALSE | color.by != "isotype"){ #if colored by something else we don't care about subtypes...
if(variant.plot == FALSE){
for (i in 1:length(sample.names)){
#subset VDJ.matrix by group.by
VDJ.matrix <- subset(VDJ.matrix.all, VDJ.matrix.all[,group.by] == sample.names[i])
#get clonotype frequency table
clono_freq <- as.data.frame(table(VDJ.matrix$clonotype_id))
clono_freq <- clono_freq[order(clono_freq$Freq, decreasing = TRUE),]
#get essential info from VDJ_GEX_matrix
curr_rep_iso <- VDJ.matrix[,c("barcode","clonotype_id", "VDJ_cgene", "VDJ_cdr3_aa", "VJ_cdr3_aa")]
names(curr_rep_iso)[3] <- "isotype"
if(color.by != "isotype"){ #add color info in case its needed
curr_rep_iso$colors <- as.character(VDJ.matrix[,color.by]) #get as character
curr_rep_iso$colors[which(is.na(curr_rep_iso$colors))] <- "None" #replace any NAs which would mess up string counting later
if(inherits(VDJ.matrix[,color.by],"factor")){ #REORDER Depending on class in VDJ.matrix. This is an attempt at conserving existing factor levels such as in the seurat_clusters column
curr_rep_iso$colors <- ordered(as.factor(curr_rep_iso$colors),levels = c(levels(VDJ.matrix[,color.by]), "None")) #reorder
} else {
curr_rep_iso$colors <- ordered(as.factor(curr_rep_iso$colors),levels = c(as.character(unique(VDJ.matrix[,color.by])), "None"))
}
}
if(treat.incomplete.clones == "exclude"){
clones_to_del <-c()
for(k in 1:nrow(clono_freq)){
if(all(curr_rep_iso[which(curr_rep_iso$clonotype_id == clono_freq[k,1]),"isotype"] == "")){ #detect if no VDJC is available in this clonotype
clones_to_del <- append(clones_to_del, k)
}
}
if(length(clones_to_del > 0)){
clono_freq <- clono_freq[-clones_to_del,]} #remove these clones entirely
clono_freq <- clono_freq[order(clono_freq$Freq, decreasing = T),] #reorder, to make sure
}
#iterate over clones and get isotype info
clones_per_isotype <- list()
for (j in 1:clones){
curr_clone <- curr_rep_iso[which(curr_rep_iso$clonotype_id == clono_freq[j,1]),]
if(nrow(curr_clone) > 0){
if(color.by == "isotype"){
#str_split and discard potentially second chain to avoid overcounting in the case of a cell having 2 HC with 2 different isotypes
curr_clone$isotype <- stringr::str_split(curr_clone$isotype, ";", simplify = TRUE)[,1]
if(treat.incomplete.cells == "proportional" & stringr::str_detect(paste0(curr_clone$isotype,collapse = ";"), pattern = "IG")==TRUE){ #check that there is at least one IG entry
props <- table(curr_clone$isotype[which(stringr::str_detect(curr_clone$isotype, pattern = "IG")==TRUE)]) #getting proportions
n_total <- nrow(curr_clone) #getting total number of cells in clonotype
props <- round(props / sum(props) * n_total,0) #calculating new number of each isotype to match proportions
if(sum(props) > n_total){props[which.max(props)] <- props[which.max(props)] - 1
} else if(sum(props) < n_total){props[which.min(props)] <- props[which.min(props)] + 1} #catching rounding derived errors.
curr_clone$isotype <- rep.int(names(props), props) #new isotype column with the new number of isotypes. ! these are not in the original order !
} else if (treat.incomplete.cells == "proportional" & stringr::str_detect(paste0(curr_clone$isotype,collapse = ";"), pattern = "IG")==FALSE){ #if no entry is present
curr_clone$isotype <- "None"
}
clones_per_isotype[[j]] <- data.frame("Counts"=rep(0, 6), "Color"=rep("", 6), "Isotype"=rep("", 6), "ClonalRank"=rep("", 6), "clonotype_id" = rep(curr_clone$clonotype_id[1],6), "VDJ_cdr3_aa" = rep(curr_clone$VDJ_cdr3_aa[which(curr_clone$VDJ_cdr3_aa != "")][1],6), "VJ_cdr3_aa" = rep(curr_clone$VJ_cdr3_aa[which(curr_clone$VJ_cdr3_aa != "")][1],6), "barcode" = rep(paste0(curr_clone$barcode, collapse = ";"),6)) #to maintain clonotype information
clones_per_isotype[[j]]$Counts[1] <- sum(stringr::str_count(curr_clone$isotype, "IGHG"))
clones_per_isotype[[j]]$Counts[2] <- sum(stringr::str_count(curr_clone$isotype, "IGHM"))
clones_per_isotype[[j]]$Counts[3] <- sum(stringr::str_count(curr_clone$isotype, "IGHA"))
clones_per_isotype[[j]]$Counts[4] <- sum(stringr::str_count(curr_clone$isotype, "IGHD"))
clones_per_isotype[[j]]$Counts[5] <- sum(stringr::str_count(curr_clone$isotype, "IGHE"))
clones_per_isotype[[j]]$Counts[6] <- sum(stringr::str_count(curr_clone$isotype, "None"))
if(color.by == "isotype"){
clones_per_isotype[[j]]$Color <- c("green4", "black", "red", "blue", "purple", "gray")
clones_per_isotype[[j]]$Isotype <- c("IGHG", "IGHM", "IGHA", "IGHD", "IGHE", "Unknown")
} else {
clones_per_isotype[[j]]$Isotype <- names(which.max(table(curr_clone$colors)))
}
clones_per_isotype[[j]]$ClonalRank <- j
} else if(color.by != "isotype"){
color_cur_clone <- unique(curr_rep_iso$colors)
n_color_cur_clone <- length(unique(curr_rep_iso$colors))
clones_per_isotype[[j]] <- data.frame("Counts"=rep(0, n_color_cur_clone), "Color"= color_cur_clone, "ClonalRank"=rep("", n_color_cur_clone), "clonotype_id" = rep(curr_clone$clonotype_id[1],n_color_cur_clone), "VDJ_cdr3_aa" = rep(curr_clone$VDJ_cdr3_aa[which(curr_clone$VDJ_cdr3_aa != "")][1],n_color_cur_clone), "VJ_cdr3_aa" = rep(curr_clone$VJ_cdr3_aa[which(curr_clone$VJ_cdr3_aa != "")][1],n_color_cur_clone), "barcode" = rep(paste0(curr_clone$barcode, collapse = ";"),n_color_cur_clone))
for(k in 1:nrow(clones_per_isotype[[j]])){
clones_per_isotype[[j]]$Counts[k] <- stringr::str_count(paste0("/",paste0(curr_clone$colors,collapse = "/ /"),"/"), pattern = paste0("/", as.character(clones_per_isotype[[j]]$Color[k]),"/"))
}
clones_per_isotype[[j]]$ClonalRank <- j
} # end not coloring by isotype but someting else
} #end if(nrow(curr_clone) > 0)
}
clones_per_isotype_all[[i]] <- do.call("rbind", clones_per_isotype)
if(treat.incomplete.cells == "exclude" & color.by == "isotype"){
rank_raw <- as.data.frame(clones_per_isotype_all[[i]] %>% dplyr::group_by(ClonalRank) %>% dplyr::summarise(sum_counts = sum(Counts)) %>% dplyr::arrange(dplyr::desc(sum_counts)) %>% dplyr::mutate(rank = 1:length(unique(ClonalRank))))
clones_per_isotype_all[[i]]$ClonalRank_2 <- 0
for(l in 1:nrow(rank_raw)){
clones_per_isotype_all[[i]]$ClonalRank_2[which(clones_per_isotype_all[[i]]$ClonalRank == rank_raw$ClonalRank[l])] <- rank_raw$rank[l]
}
clones_per_isotype_all[[i]]$ClonalRank <- clones_per_isotype_all[[i]]$ClonalRank_2
#reorder the dataframe the sum_counts order by decreasing sum, the clonal rank is there to keep clones that have the same count sum together as a group of rows
message("New ranking based only on present VDJ chains: ")
message(unique(clones_per_isotype_all[[i]]$clonotype_id))
}
#Delete clones not of interest
if(color.by == "isotype" & isotypes.to.plot[1] != "all"){
if(!any(isotypes.to.plot %in% unique(clones_per_isotype_all[[i]]$Isotype))){stop("isotype.to.plot input not found in dataframe. Please check if the isotype is spelled correctly")}
to_del <- c()
for(k in 1:length(unique(clones_per_isotype_all[[i]]$ClonalRank))){
if(sum(clones_per_isotype_all[[i]]$Counts[which(clones_per_isotype_all[[i]]$ClonalRank == unique(clones_per_isotype_all[[i]]$ClonalRank)[k] & clones_per_isotype_all[[i]]$Isotype %in% isotypes.to.plot)]) == 0){
to_del <- append(to_del, unique(clones_per_isotype_all[[i]]$ClonalRank)[k])
}
}
clones_per_isotype_all[[i]] <- subset(clones_per_isotype_all[[i]], !ClonalRank %in% to_del)
#redistribute clonal ranks
rank_raw <- as.data.frame(clones_per_isotype_all[[i]] %>% dplyr::group_by(ClonalRank) %>% dplyr::summarise(sum_counts = sum(Counts)) %>% dplyr::arrange(dplyr::desc(sum_counts)) %>% dplyr::mutate(rank = 1:length(unique(ClonalRank))))
clones_per_isotype_all[[i]]$ClonalRank_2 <- 0
for(l in 1:nrow(rank_raw)){
clones_per_isotype_all[[i]]$ClonalRank_2[which(clones_per_isotype_all[[i]]$ClonalRank == rank_raw$ClonalRank[l])] <- rank_raw$rank[l]
}
clones_per_isotype_all[[i]]$ClonalRank <- clones_per_isotype_all[[i]]$ClonalRank_2
message("New ranking based only on selected isotypes: ")
message(unique(clones_per_isotype_all[[i]]$clonotype_id))
}
if(color.by == "isotype"){
output_plot[[i]] <- ggplot2::ggplot(clones_per_isotype_all[[i]], ggplot2::aes(fill = Isotype, y=Counts, x=ClonalRank)) +
ggplot2::geom_bar(stat="identity", width=0.6, color="black") +
ggplot2::theme_bw() +
ggplot2::scale_fill_manual(values = ggplot2::scale_fill_manual(values = c("IGHG" = "green4", "IGHM" = "black", "IGHA" = "red3", "IGHD"="blue", "IGHE"="purple", "Unknown"="gray"))) +
ggplot2::theme_classic() +
ggplot2::ggtitle(paste0(i)) +
ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5), text = ggplot2::element_text(size = text.size)) +
ggplot2::scale_y_continuous(expand = c(0,0)) + ggplot2::scale_x_continuous(expand = c(0,0.5)) +
ggplot2::labs(title = sample.names[[i]], x = "Clonal rank", y = "Number of cells")
} else {
output_plot[[i]] <- ggplot2::ggplot(clones_per_isotype_all[[i]], ggplot2::aes(fill = Color, y=Counts, x=ClonalRank)) + ggplot2::geom_bar(stat="identity", width=0.6, color="black") + ggplot2::theme_bw() + ggplot2::theme_classic() + ggplot2::ggtitle(paste0(i)) + ggplot2::theme(text = ggplot2::element_text(size = text.size), plot.title = ggplot2::element_text(hjust = 0.5)) + ggplot2::scale_y_continuous(expand = c(0,0)) + ggplot2::scale_x_continuous(expand = c(0,0.5)) + ggplot2::labs(title = sample.names[[i]], x = "Clonal rank", y = "Number of cells", fill = color.by) + ggplot2::scale_fill_manual(values = unique_colors,labels=names(unique_colors))
}
}
names(clones_per_isotype_all) <- sample.names
return(list(output_plot,clones_per_isotype_all))
}
if(variant.plot ==TRUE){
for (i in 1:length(sample.names)){
message(paste0("Starting sample ", i, "/", length(sample.names)))
#subset VDJ.matrix by group.by
VDJ.matrix <- subset(VDJ.matrix.all, VDJ.matrix.all[,group.by] == sample.names[i])
#Rank clonotypes
s_id = VDJ.matrix$clonotype_id
s_id_num = readr::parse_number(s_id)
s_id_num = sprintf("%05d", s_id_num)
s_id_num = paste0("clonotype", s_id_num)
VDJ.matrix$clonotype_id <- s_id_num
s_id = NULL
s_id_num = NULL
VDJ.matrix$clonotype_id = paste0("old", VDJ.matrix$clonotype_id)
rank_raw <- as.data.frame(VDJ.matrix %>% dplyr::group_by(clonotype_id) %>% dplyr::summarise(n = n()))
rank_raw <- rank_raw[order(rank_raw$n, rank_raw$clonotype_id, decreasing = TRUE),]
#rank_raw <- rank_raw[order(rank_raw$n, decreasing = T),]
rank_raw$new_clonotype <- 1:length(rank_raw$clonotype_id)
rank_raw$new_clonotype = sprintf("%05d", rank_raw$new_clonotype)
rank_raw$new_clonotype = paste0("new", rank_raw$new_clonotype)
for(z in 1:nrow(rank_raw)){
VDJ.matrix$clonotype_id <- gsub(rank_raw$clonotype_id[z], rank_raw$new_clonotype[z], VDJ.matrix$clonotype_id) #For VDJ
}
VDJ.matrix$clonotype_id <- paste0("clonotype", sprintf("%05d", readr::parse_number(VDJ.matrix$clonotype_id)))
#get clonotype frequency table
clono_freq <- as.data.frame(table(VDJ.matrix$clonotype_id))
clono_freq <- clono_freq[order(clono_freq$Freq, decreasing = TRUE),]
is.trimmed = VDJ.matrix$VJ_sequence_nt_trimmed
#Define the variants
if (is.null(is.trimmed) == FALSE){
VDJ.matrix$pasted_variants = paste(VDJ.matrix$VDJ_sequence_nt_trimmed, VDJ.matrix$VJ_sequence_nt_trimmed, sep = ";")
message(paste0("Trimmed sequences found for sample ", sample.names[i]))
message("Variants are obtained as VDJ_sequence_nt_trimmed;VJ_sequence_nt_trimmed")
}
if (is.null(is.trimmed) == TRUE){
VDJ.matrix$pasted_variants = paste(VDJ.matrix$VDJ_sequence_nt_raw, VDJ.matrix$VJ_sequence_nt_raw, sep = ";")
message(paste0("Trimmed sequences not found for sample ", sample.names[i]))
message("Variants are obtained as VDJ_sequence_nt_raw;VJ_sequence_nt_raw")
}
#get essential info from VDJ_GEX_matrix
curr_rep_iso <- VDJ.matrix[,c("barcode","clonotype_id", "VDJ_cgene", "pasted_variants")]
names(curr_rep_iso)[3] <- "isotype"
#We only care about the main isotypes, not the subgroups
curr_rep_iso$isotype <- substr(curr_rep_iso$isotype,1,4)
if(color.by != "isotype"){ #add color info in case its needed
curr_rep_iso$colors <- as.character(VDJ.matrix[,color.by]) #get as character
curr_rep_iso$colors[which(is.na(curr_rep_iso$colors))] <- "None" #replace any NAs which would mess up string counting later
if(inherits(VDJ.matrix[,color.by],"factor")){ #REORDER Depending on class in VDJ.matrix. This is an attempt at conserving existing factor levels such as in the seurat_clusters column
curr_rep_iso$colors <- ordered(as.factor(curr_rep_iso$colors),levels = c(levels(VDJ.matrix[,color.by]), "None")) #reorder
} else {
curr_rep_iso$colors <- ordered(as.factor(curr_rep_iso$colors),levels = c(as.character(unique(VDJ.matrix[,color.by])), "None"))
}
}
if(treat.incomplete.clones == "exclude"){
clones_to_del <-c()
for(k in 1:nrow(clono_freq)){
if(all(curr_rep_iso[which(curr_rep_iso$clonotype_id == clono_freq[k,1]),"isotype"] == "")){ #detect if no VDJC is available in this clonotype
clones_to_del <- append(clones_to_del, k)
}
}
if(length(clones_to_del > 0)){
#Formatting numbers inside the function
s_id_num = clones_to_del
s_id_num = sprintf("%05d", s_id_num)
s_id_num = paste0("clonotype", s_id_num)
clones_to_del <- s_id_num
s_id = NULL
s_id_num = NULL
s_id = curr_rep_iso$clonotype_id
s_id_num = readr::parse_number(s_id)
s_id_num = sprintf("%05d", s_id_num)
s_id_num = paste0("clonotype", s_id_num)
curr_rep_iso$clonotype_id <- s_id_num
s_id = NULL
s_id_num = NULL
#Delete the clones that are incomplete
curr_rep_iso$match = curr_rep_iso$clonotype_id %in% clones_to_del
curr_rep_iso <- subset(curr_rep_iso, subset = match == FALSE)
curr_rep_iso$match = NULL
curr_rep_iso= curr_rep_iso[order(curr_rep_iso$clonotype_id),]
curr_rep_iso$clonotype_id <- paste0("old", curr_rep_iso$clonotype_id)
#Re rank the clonotypes
rank_raw <- as.data.frame(curr_rep_iso %>% dplyr::group_by(clonotype_id) %>% dplyr::summarise(n = n()))
rank_raw <- rank_raw[order(rank_raw$n, rank_raw$clonotype_id, decreasing = TRUE),]
rank_raw$new_clonotype <- 1:length(rank_raw$clonotype_id)
rank_raw$new_clonotype = paste0("newclonotype", sprintf("%05d", rank_raw$new_clonotype))
#Replace in the dataframe the old clonotypes by the new rank
for(z in 1:nrow(rank_raw)){
curr_rep_iso$clonotype_id <- gsub(rank_raw$clonotype_id[z], rank_raw$new_clonotype[z], curr_rep_iso$clonotype_id) #For VDJ
}
curr_rep_iso <- curr_rep_iso[order(curr_rep_iso$clonotype_id),]
}
}
#Formatting the clonotypes for the function
s_id = curr_rep_iso$clonotype_id
s_id_num = readr::parse_number(s_id)
s_id_num = sprintf("%05d", s_id_num)
s_id_num = paste0("clonotype", s_id_num)
curr_rep_iso$clonotype_id <- s_id_num
s_id = NULL
s_id_num = NULL
options(dplyr.summarise.inform = FALSE)
#Obtain the variant df depending on coloring
if(color.by == "isotype"){
curr_rep_iso %>% dplyr::group_by(pasted_variants, clonotype_id, isotype) %>% dplyr::summarise(n=n()) -> variant_df
}
if(color.by != "isotype"){
curr_rep_iso %>% dplyr::group_by(pasted_variants, clonotype_id, colors) %>% dplyr::summarise(n=n()) -> variant_df
}
variant_df= variant_df[order(variant_df$clonotype_id,variant_df$n),]
#Simplify the variant_df to keep only the number of clones desired
max_row = max(which(variant_df$clonotype_id == paste0("clonotype", sprintf("%05d", clones))))
variant_df = variant_df[1:max_row,]
s_id = variant_df$clonotype_id
s_id_num = readr::parse_number(s_id)
variant_df$clonotype_id <- s_id_num
s_id = NULL
s_id_num = NULL
if(treat.incomplete.cells == "proportional" & stringr::str_detect(paste0(variant_df$isotype,collapse = ";"), pattern = "IG")==TRUE){ #check that there is at least one IG entry
for (clone_number in 1:clones){
subset_clone_variant_with_isotype = variant_df[which(variant_df$clonotype_id == clone_number & stringr::str_detect(variant_df$isotype, pattern = "IG")==TRUE),]
subset_clone_variant = variant_df[which(variant_df$clonotype_id == clone_number),]
variants <- c()
if ((nrow(subset_clone_variant_with_isotype) < nrow(subset_clone_variant)) & (nrow(subset_clone_variant_with_isotype) >= 1)){
for (z in 1:nrow(subset_clone_variant_with_isotype)){
variants <- c(variants, rep(subset_clone_variant_with_isotype$isotype[z], subset_clone_variant_with_isotype$n[z]))
}
proportional_variants = c()
for (variantes in 1:length(which(variant_df$clonotype_id == clone_number & stringr::str_detect(variant_df$isotype, pattern = "IG")==FALSE))){
proportional_variants= c(proportional_variants, sample(variants, 1))
}
variant_df[which(variant_df$clonotype_id == clone_number & stringr::str_detect(variant_df$isotype, pattern = "IG")==FALSE),]$isotype = proportional_variants
} else if ((nrow(subset_clone_variant_with_isotype) < nrow(subset_clone_variant)) & (nrow(subset_clone_variant_with_isotype) == 0)){
proportional_variants = c()
proportional_variants = c(proportional_variants, rep("Unknown", nrow(subset_clone_variant)))
variant_df[which(variant_df$clonotype_id == clone_number & stringr::str_detect(variant_df$isotype, pattern = "IG")==FALSE),]$isotype = proportional_variants
}
}
}
if(treat.incomplete.cells == "exclude" & color.by == "isotype"){
variant_df$match_ex_crit <- variant_df$isotype == ""
variant_df <- subset(variant_df, subset = match_ex_crit == FALSE)
variant_df$match_ex_crit = NULL
variant_df$original_clonotypes = variant_df$clonotype_id
variant_df$clonotype_id = sprintf("%05d", variant_df$clonotype_id)
variant_df$clonotype_id = paste0("old", variant_df$clonotype_id)
rank_raw <- as.data.frame(variant_df %>% dplyr::group_by(clonotype_id) %>% dplyr::summarise(n = sum(n)))
rank_raw <- rank_raw[order(rank_raw$n, rank_raw$clonotype_id, decreasing = TRUE),]
rank_raw$new_clonotype <- 1:length(rank_raw$clonotype_id)
rank_raw$new_clonotype = sprintf("%05d", rank_raw$new_clonotype)
rank_raw$new_clonotype = paste0("new", rank_raw$new_clonotype)
for(z in 1:nrow(rank_raw)){
variant_df$clonotype_id <- gsub(rank_raw$clonotype_id[z], rank_raw$new_clonotype[z], variant_df$clonotype_id) #For VDJ
}
variant_df$clonotype_id <- readr::parse_number(variant_df$clonotype_id)
variant_df <- variant_df[order(variant_df$clonotype_id),]
}
#Delete clones not of interest
if(color.by == "isotype" & isotypes.to.plot[1] != "all"){
if(!any(isotypes.to.plot %in% unique(variant_df$isotype))){stop("isotype.to.plot input not found in dataframe. Please check if the isotype is spelled correctly")}
variant_df$match_ex_crit <- variant_df$isotype %in% isotypes.to.plot
variant_df <- subset(variant_df, subset = match_ex_crit == TRUE)
variant_df$match_ex_crit = NULL
variant_df$original_clonotypes = variant_df$clonotype_id
variant_df$clonotype_id = sprintf("%05d", variant_df$clonotype_id)
variant_df$clonotype_id = paste0("old", variant_df$clonotype_id)
rank_raw <- as.data.frame(variant_df %>% dplyr::group_by(clonotype_id) %>% dplyr::summarise(n = sum(n)))
rank_raw <- rank_raw[order(rank_raw$n, rank_raw$clonotype_id, decreasing = TRUE),]
rank_raw$new_clonotype <- 1:length(rank_raw$clonotype_id)
rank_raw$new_clonotype = sprintf("%05d", rank_raw$new_clonotype)
rank_raw$new_clonotype = paste0("new", rank_raw$new_clonotype)
for(z in 1:nrow(rank_raw)){
variant_df$clonotype_id <- gsub(rank_raw$clonotype_id[z], rank_raw$new_clonotype[z], variant_df$clonotype_id) #For VDJ
}
variant_df$clonotype_id <- readr::parse_number(variant_df$clonotype_id)
variant_df <- variant_df[order(variant_df$clonotype_id),]
message("New ranking based only on selected isotypes: ")
message(paste0(unique(variant_df$clonotype_id),", "))
}
variant_df$variant = 1:length(variant_df$clonotype_id)
variant_df$variant = as.character(variant_df$variant)
variant_df_list[[i]] = variant_df
if(color.by == "isotype"){
output_plot[[i]] <- ggplot2::ggplot(variant_df_list[[i]], ggplot2::aes(x = clonotype_id, y = n, fill = isotype, color = variant)) + ggplot2::geom_bar(stat="identity", width=0.6, color="white") + ggplot2::theme_bw() + ggplot2::scale_fill_manual("Isotype",values = c("IGHG" = "green4", "IGHM" = "black", "IGHA" = "red3", "IGHD"="blue", "IGHE"="purple", "Unknown"="gray")) + ggplot2::theme_classic() + ggplot2::ggtitle(paste0(i)) + ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5)) + ggplot2::scale_y_continuous(expand = c(0,0)) + ggplot2::scale_x_continuous(expand = c(0,0.5)) + ggplot2::labs(title = sample.names[[i]], x = "Clonal rank", y = "Number of cells")
} else {
output_plot[[i]] <- ggplot2::ggplot(variant_df_list[[i]], ggplot2::aes(x = clonotype_id, y = n, fill = colors, color = variant)) + ggplot2::geom_bar(stat="identity", width=0.6, color="white") + ggplot2::scale_fill_manual(values = grDevices::rainbow(n = length(unique(clones_per_isotype_all[[i]]$colors)))) + ggplot2::theme_classic() + ggplot2::ggtitle(paste0(i)) + ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5)) + ggplot2::scale_y_continuous(expand = c(0,0)) + ggplot2::scale_x_continuous(expand = c(0,0.5)) + ggplot2::labs(title = sample.names[[i]], x = "Clonal rank", y = "Number of cells")
}
}
clones_per_isotype_all = variant_df_list
names(clones_per_isotype_all) <- sample.names
names(output_plot) <- sample.names
return(list(output_plot,clones_per_isotype_all))
}
}
if(subtypes ==TRUE & color.by == "isotype"){
if(variant.plot == FALSE){
for (i in 1:length(sample.names)){
#subset VDJ.matrix by group.by
VDJ.matrix <- subset(VDJ.matrix.all, VDJ.matrix.all[,group.by] == sample.names[i])
#get clonotype frequency table
clono_freq <- as.data.frame(table(VDJ.matrix$clonotype_id))
clono_freq <- clono_freq[order(clono_freq$Freq, decreasing = TRUE),]
#get essential info from VDJ_GEX_matrix
curr_rep_iso <- VDJ.matrix[,c("barcode","clonotype_id", "VDJ_cgene", "VDJ_cdr3_aa", "VJ_cdr3_aa")]
curr_rep_iso$VDJ_cgene[is.null(curr_rep_iso$VDJ_cgene)] <- "none"
names(curr_rep_iso)[3] <- "isotype"
#no substring here, just splitting on a ; to exclude multiple isotypes of one clone
curr_rep_iso$isotype <- stringr::str_split(curr_rep_iso$isotype, ";", simplify = TRUE)[,1]
if(treat.incomplete.clones == "exclude"){
clones_to_del <-c()
for(k in 1:nrow(clono_freq)){
if(all(curr_rep_iso[which(curr_rep_iso$clonotype_id == clono_freq[k,1]),"isotype"] == "")){ #detect if no VDJC is available in this clonotype
clones_to_del <- append(clones_to_del, k)
}
}
if(length(clones_to_del > 0)){
clono_freq <- clono_freq[-clones_to_del,]} #remove these clones entirely
clono_freq <- clono_freq[order(clono_freq$Freq, decreasing = TRUE),] #reorder, to make sure
}
#iterate over clones and get isotype info
clones_per_isotype <- list()
j <- 1
for (j in 1:clones){
curr_clone <- curr_rep_iso[which(curr_rep_iso$clonotype_id == clono_freq[j,1]),]
if(nrow(curr_clone) > 0){
#str_split and discard potentially second chain to avoid overcounting in the case of a cell having 2 HC with 2 different isotypes
curr_clone$isotype <- stringr::str_split(curr_clone$isotype, ";", simplify = TRUE)[,1]
if(treat.incomplete.cells == "proportional" & stringr::str_detect(paste0(curr_clone$isotype,collapse = ";"), pattern = "IG")==TRUE){ #check that there is at least one IG entry
props <- table(curr_clone$isotype[which(stringr::str_detect(curr_clone$isotype, pattern = "IG")==TRUE)]) #getting proportions
n_total <- nrow(curr_clone) #getting total number of cells in clonotype
props <- round(props / sum(props) * n_total,0) #calculating new number of each isotype to match proportions
if(sum(props) > n_total){props[which.max(props)] <- props[which.max(props)] - 1
} else if(sum(props) < n_total){props[which.min(props)] <- props[which.min(props)] + 1} #catching rounding derived errors.
curr_clone$isotype <- rep.int(names(props), props) #new isotype column with the new number of isotypes. ! these are not in the original order !
} else if (treat.incomplete.cells == "proportional" & stringr::str_detect(paste0(curr_clone$isotype,collapse = ";"), pattern = "IG")==FALSE){ #if no entry is present
curr_clone$isotype <- "None"
}
clones_per_isotype[[j]] <- data.frame("Counts"=rep(0, 14), "Color"=rep("", 14), "Isotype"=rep("", 14), "ClonalRank"=rep("", 14), "clonotype_id" = rep(curr_clone$clonotype_id[1],14), "VDJ_cdr3_aa" = rep(curr_clone$VDJ_cdr3_aa[which(curr_clone$VDJ_cdr3_aa != "")][1],14), "VJ_cdr3_aa" = rep(curr_clone$VJ_cdr3_aa[which(curr_clone$VJ_cdr3_aa != "")][1],14), "barcode" = rep(paste0(curr_clone$barcode, collapse = ";"),14)) #to maintain clonotype information
clones_per_isotype[[j]]$Counts[1] <- sum(stringr::str_count(curr_clone$isotype, "IGHG1"))
clones_per_isotype[[j]]$Counts[2] <- sum(stringr::str_count(curr_clone$isotype, "IGHG2"))
clones_per_isotype[[j]]$Counts[3] <- sum(stringr::str_count(curr_clone$isotype, "IGHG2A"))
clones_per_isotype[[j]]$Counts[4] <- sum(stringr::str_count(curr_clone$isotype, "IGHG2B"))
clones_per_isotype[[j]]$Counts[5] <- sum(stringr::str_count(curr_clone$isotype, "IGHG2C"))
clones_per_isotype[[j]]$Counts[6] <- sum(stringr::str_count(curr_clone$isotype, "IGHG3"))
clones_per_isotype[[j]]$Counts[7] <- sum(stringr::str_count(curr_clone$isotype, "IGHG4"))
clones_per_isotype[[j]]$Counts[8] <- sum(stringr::str_count(curr_clone$isotype, "IGHM"))
clones_per_isotype[[j]]$Counts[9] <- sum(stringr::str_count(curr_clone$isotype, "IGHA"))
clones_per_isotype[[j]]$Counts[10] <- sum(stringr::str_count(curr_clone$isotype, "IGHA1"))
clones_per_isotype[[j]]$Counts[11] <- sum(stringr::str_count(curr_clone$isotype, "IGHA2"))
clones_per_isotype[[j]]$Counts[12] <- sum(stringr::str_count(curr_clone$isotype, "IGHD"))
clones_per_isotype[[j]]$Counts[13] <- sum(stringr::str_count(curr_clone$isotype, "IGHE"))
clones_per_isotype[[j]]$Counts[14] <- sum(stringr::str_count(curr_clone$isotype, "None"))
clones_per_isotype[[j]]$ClonalRank <- j
if(color.by == "isotype"){
clones_per_isotype[[j]]$Isotype <- c("IGHG1", "IGHG2", "IGHG2a", "IGHG2b", "IGHG2c", "IGHG3", "IGHG4", "IGHM", "IGHA", "IGHA1", "IGHA2", "IGHD", "IGHE", "Unknown")
if (species == "Human"){
clones_per_isotype[[j]] <- clones_per_isotype[[j]][-c(3, 4, 5, 9), ]
}
if (species == "Mouse"){
clones_per_isotype[[j]] <- clones_per_isotype[[j]][-c(2, 7, 10, 11), ]
}
} else {
clones_per_isotype[[j]]$Isotype <- names(which.max(table(curr_clone$colors)))
}
} #end if(nrow(curr_clone) > 0)
}
clones_per_isotype_all[[i]] <- do.call("rbind",clones_per_isotype)
if(treat.incomplete.cells == "exclude"){
rank_raw <- as.data.frame(clones_per_isotype_all[[i]] %>% dplyr::group_by(ClonalRank) %>% dplyr::summarise(sum_counts = sum(Counts)) %>% dplyr::arrange(dplyr::desc(sum_counts)) %>% dplyr::mutate(rank = 1:length(unique(ClonalRank))))
clones_per_isotype_all[[i]]$ClonalRank_2 <- 0
for(l in 1:nrow(rank_raw)){
clones_per_isotype_all[[i]]$ClonalRank_2[which(clones_per_isotype_all[[i]]$ClonalRank == rank_raw$ClonalRank[l])] <- rank_raw$rank[l]
}
clones_per_isotype_all[[i]]$ClonalRank <- clones_per_isotype_all[[i]]$ClonalRank_2
#reorder the dataframe the sum_counts order by decreasing sum, the clonal rank is there to keep clones that have the same count sum together as a group of rows
message("New ranking based only on present HC chains: ")
message(unique(clones_per_isotype_all[[i]]$clonotype_id))
}
#Delete clones not of interest
if(color.by == "isotype" & isotypes.to.plot[1] != "all"){
if(!any(isotypes.to.plot %in% unique(clones_per_isotype_all[[i]]$Isotype))){stop("isotype.to.plot input not found in dataframe. Please check if the isotype is spelled correctly")}
to_del <- c()
for(k in 1:length(unique(clones_per_isotype_all[[i]]$ClonalRank))){
if(sum(clones_per_isotype_all[[i]]$Counts[which(clones_per_isotype_all[[i]]$ClonalRank == unique(clones_per_isotype_all[[i]]$ClonalRank)[k] & clones_per_isotype_all[[i]]$Isotype %in% isotypes.to.plot)]) == 0){
to_del <- append(to_del, unique(clones_per_isotype_all[[i]]$ClonalRank)[k])
}
}
clones_per_isotype_all[[i]] <- subset(clones_per_isotype_all[[i]], !ClonalRank %in% to_del)
#redistribute clonal ranks
rank_raw <- as.data.frame(clones_per_isotype_all[[i]] %>% dplyr::group_by(ClonalRank) %>% dplyr::summarise(sum_counts = sum(Counts)) %>% plyr::arrange(dplyr::desc(sum_counts)) %>% dplyr::mutate(rank = 1:length(unique(ClonalRank))))
clones_per_isotype_all[[i]]$ClonalRank_2 <- 0
for(l in 1:nrow(rank_raw)){
clones_per_isotype_all[[i]]$ClonalRank_2[which(clones_per_isotype_all[[i]]$ClonalRank == rank_raw$ClonalRank[l])] <- rank_raw$rank[l]
}
clones_per_isotype_all[[i]]$ClonalRank <- clones_per_isotype_all[[i]]$ClonalRank_2
message("New ranking based only on selected isotypes: ")
message(unique(clones_per_isotype_all[[i]]$clonotype_id))
}
if (species == "Human"){
output_plot[[i]] <- ggplot2::ggplot(clones_per_isotype_all[[i]], ggplot2::aes(fill = Isotype, y=Counts, x=ClonalRank)) + ggplot2::geom_bar(stat="identity", width=0.6, color="black") + ggplot2::theme_bw() + ggplot2::scale_fill_manual("Isotype", values = c("IGHG1" = "green","IGHG2" = "green3", "IGHG3"="green4", "IGHG4"="darkgreen", "IGHM" = "black", "IGHA1" = "red", "IGHA2"= "red4", "IGHD"="blue", "IGHE"="purple", "Unknown"="gray")) + ggplot2::theme_classic() + ggplot2::ggtitle(sample.names[i]) + 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_x_continuous(expand = c(0,0.5))
}
if (species == "Mouse"){
output_plot[[i]] <- ggplot2::ggplot(clones_per_isotype_all[[i]], ggplot2::aes(fill = Isotype, y=Counts, x=ClonalRank)) + ggplot2::geom_bar(stat="identity", width=0.6, color="black") + ggplot2::theme_bw() + ggplot2::scale_fill_manual("Isotype", values = c("IGHG1" = "lightgreen", "IGHG2a"="green", "IGHG2b" = "green3", "IGHG2c"="green4", "IGHG3"="darkgreen", "IGHM" = "black", "IGHA" = "red", "IGHD"="blue", "IGHE"="purple", "Unknown"="gray")) + ggplot2::theme_classic() + ggplot2::ggtitle(sample.names[i]) + 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_x_continuous(expand = c(0,0.5))
}
}
names(clones_per_isotype_all) <- sample.names
return(list(output_plot,clones_per_isotype_all))
}
if(variant.plot ==TRUE){
for (i in 1:length(sample.names)){
message(paste0("Starting sample ", i, "/", length(sample.names)))
#subset VDJ.matrix by group.by
VDJ.matrix <- subset(VDJ.matrix.all, VDJ.matrix.all[,group.by] == sample.names[i])
#Rank clonotypes
s_id = VDJ.matrix$clonotype_id
s_id_num = readr::parse_number(s_id)
s_id_num = sprintf("%05d", s_id_num)
s_id_num = paste0("clonotype", s_id_num)
VDJ.matrix$clonotype_id <- s_id_num
s_id = NULL
s_id_num = NULL
VDJ.matrix$clonotype_id = paste0("old", VDJ.matrix$clonotype_id)
rank_raw <- as.data.frame(VDJ.matrix %>% dplyr::group_by(clonotype_id) %>% dplyr::summarise(n = n()))
rank_raw <- rank_raw[order(rank_raw$n, rank_raw$clonotype_id, decreasing = T),]
rank_raw$new_clonotype <- 1:length(rank_raw$clonotype_id)
rank_raw$new_clonotype = sprintf("%05d", rank_raw$new_clonotype)
rank_raw$new_clonotype = paste0("new", rank_raw$new_clonotype)
for(z in 1:nrow(rank_raw)){
VDJ.matrix$clonotype_id <- gsub(rank_raw$clonotype_id[z], rank_raw$new_clonotype[z], VDJ.matrix$clonotype_id) #For VDJ
}
VDJ.matrix$clonotype_id <- paste0("clonotype", sprintf("%05d", readr::parse_number(VDJ.matrix$clonotype_id)))
#get clonotype frequency table
clono_freq <- as.data.frame(table(VDJ.matrix$clonotype_id))
clono_freq <- clono_freq[order(clono_freq$Freq, decreasing = TRUE),]
is.trimmed = VDJ.matrix$VJ_sequence_nt_trimmed
#Define the variants
if (is.null(is.trimmed) == FALSE){
VDJ.matrix$pasted_variants = paste(VDJ.matrix$VDJ_sequence_nt_trimmed, VDJ.matrix$VJ_sequence_nt_trimmed, sep = ";")
message(paste0("Trimmed sequences found for sample ", sample.names[i]))
message("Variants are obtained as VDJ_sequence_nt_trimmed;VJ_sequence_nt_trimmed")
}
if (is.null(is.trimmed) == TRUE){
VDJ.matrix$pasted_variants = paste(VDJ.matrix$VDJ_sequence_nt_raw, VDJ.matrix$VJ_sequence_nt_raw, sep = ";")
message(paste0("Trimmed sequences not found for sample ", sample.names[i]))
message("Variants are obtained as VDJ_sequence_nt_raw;VJ_sequence_nt_raw")
}
#get essential info from VDJ_GEX_matrix
#get essential info from VDJ_GEX_matrix
curr_rep_iso <- VDJ.matrix[,c("barcode","clonotype_id", "VDJ_cgene", "pasted_variants")]
names(curr_rep_iso)[3] <- "isotype"
#no substring here, just splitting on a ; to exclude multiple isotypes of one clone
curr_rep_iso$isotype <- stringr::str_split(curr_rep_iso$isotype, ";", simplify = TRUE)[,1]
if(treat.incomplete.clones == "exclude"){
clones_to_del <-c()
for(k in 1:nrow(clono_freq)){
if(all(curr_rep_iso[which(curr_rep_iso$clonotype_id == clono_freq[k,1]),"isotype"] == "")){ #detect if no VDJC is available in this clonotype
clones_to_del <- append(clones_to_del, k)
}
}
if(length(clones_to_del > 0)){
#Formatting numbers inside the function
s_id_num = clones_to_del
s_id_num = sprintf("%05d", s_id_num)
s_id_num = paste0("clonotype", s_id_num)
clones_to_del <- s_id_num
s_id = NULL
s_id_num = NULL
s_id = curr_rep_iso$clonotype_id
s_id_num = readr::parse_number(s_id)
s_id_num = sprintf("%05d", s_id_num)
s_id_num = paste0("clonotype", s_id_num)
curr_rep_iso$clonotype_id <- s_id_num
s_id = NULL
s_id_num = NULL
#Delete the clones that are incomplete
curr_rep_iso$match = curr_rep_iso$clonotype_id %in% clones_to_del
curr_rep_iso <- subset(curr_rep_iso, subset = match == FALSE)
curr_rep_iso$match = NULL
curr_rep_iso= curr_rep_iso[order(curr_rep_iso$clonotype_id),]
curr_rep_iso$clonotype_id <- paste0("old", curr_rep_iso$clonotype_id)
#Re rank the clonotypes
rank_raw <- as.data.frame(curr_rep_iso %>% dplyr::group_by(clonotype_id) %>% dplyr::summarise(n = n()))
rank_raw <- rank_raw[order(rank_raw$n, rank_raw$clonotype_id, decreasing = TRUE),]
rank_raw$new_clonotype <- 1:length(rank_raw$clonotype_id)
rank_raw$new_clonotype = paste0("newclonotype", sprintf("%05d", rank_raw$new_clonotype))
#Replace in the dataframe the old clonotypes by the new rank
for(z in 1:nrow(rank_raw)){
curr_rep_iso$clonotype_id <- gsub(rank_raw$clonotype_id[z], rank_raw$new_clonotype[z], curr_rep_iso$clonotype_id) #For VDJ
}
curr_rep_iso <- curr_rep_iso[order(curr_rep_iso$clonotype_id),]
}
}
#Formatting the clonotypes for the function
s_id = curr_rep_iso$clonotype_id
s_id_num = readr::parse_number(s_id)
s_id_num = sprintf("%05d", s_id_num)
s_id_num = paste0("clonotype", s_id_num)
curr_rep_iso$clonotype_id <- s_id_num
s_id = NULL
s_id_num = NULL
options(dplyr.summarise.inform = FALSE)
curr_rep_iso %>% dplyr::group_by(pasted_variants, clonotype_id, isotype) %>% dplyr::summarise(n=n()) -> variant_df
variant_df= variant_df[order(variant_df$clonotype_id,variant_df$n),]
#Simplify the variant_df to keep only the number of clones desired
max_row = max(which(variant_df$clonotype_id == paste0("clonotype", sprintf("%05d", clones))))
variant_df = variant_df[1:max_row,]
s_id = variant_df$clonotype_id
s_id_num = readr::parse_number(s_id)
variant_df$clonotype_id <- s_id_num
s_id = NULL
s_id_num = NULL
if(treat.incomplete.cells == "proportional" & stringr::str_detect(paste0(variant_df$isotype,collapse = ";"), pattern = "IG")==TRUE){ #check that there is at least one IG entry
for (clone_number in 1:clones){
subset_clone_variant_with_isotype = variant_df[which(variant_df$clonotype_id == clone_number & stringr::str_detect(variant_df$isotype, pattern = "IG")==TRUE),]
subset_clone_variant = variant_df[which(variant_df$clonotype_id == clone_number),]
variants <- c()
if ((nrow(subset_clone_variant_with_isotype) < nrow(subset_clone_variant)) & (nrow(subset_clone_variant_with_isotype) >= 1)){
for (z in 1:nrow(subset_clone_variant_with_isotype)){
variants <- c(variants, rep(subset_clone_variant_with_isotype$isotype[z], subset_clone_variant_with_isotype$n[z]))
}
proportional_variants = c()
for (variantes in 1:length(which(variant_df$clonotype_id == clone_number & stringr::str_detect(variant_df$isotype, pattern = "IG")==FALSE))){
proportional_variants= c(proportional_variants, sample(variants, 1))
}
variant_df[which(variant_df$clonotype_id == clone_number & stringr::str_detect(variant_df$isotype, pattern = "IG")==FALSE),]$isotype = proportional_variants
} else if ((nrow(subset_clone_variant_with_isotype) < nrow(subset_clone_variant)) & (nrow(subset_clone_variant_with_isotype) == 0)){
proportional_variants = c()
proportional_variants = c(proportional_variants, rep("Unknown", nrow(subset_clone_variant)))
variant_df[which(variant_df$clonotype_id == clone_number & stringr::str_detect(variant_df$isotype, pattern = "IG")==FALSE),]$isotype = proportional_variants
}
}
}
if(treat.incomplete.cells == "exclude" & color.by == "isotype"){
variant_df$match_ex_crit <- variant_df$isotype == ""
variant_df <- subset(variant_df, subset = match_ex_crit == FALSE)
variant_df$match_ex_crit = NULL
variant_df$original_clonotypes = variant_df$clonotype_id
variant_df$clonotype_id = sprintf("%05d", variant_df$clonotype_id)
variant_df$clonotype_id = paste0("old", variant_df$clonotype_id)
rank_raw <- as.data.frame(variant_df %>% dplyr::group_by(clonotype_id) %>% dplyr::summarise(n = sum(n)))
rank_raw <- rank_raw[order(rank_raw$n, rank_raw$clonotype_id, decreasing = TRUE),]
rank_raw$new_clonotype <- 1:length(rank_raw$clonotype_id)
rank_raw$new_clonotype = sprintf("%05d", rank_raw$new_clonotype)
rank_raw$new_clonotype = paste0("new", rank_raw$new_clonotype)
for(z in 1:nrow(rank_raw)){
variant_df$clonotype_id <- gsub(rank_raw$clonotype_id[z], rank_raw$new_clonotype[z], variant_df$clonotype_id) #For VDJ
}
variant_df$clonotype_id <- readr::parse_number(variant_df$clonotype_id)
variant_df <- variant_df[order(variant_df$clonotype_id),]
}
if(color.by == "isotype" & isotypes.to.plot[1] != "all"){
if(!any(isotypes.to.plot %in% unique(variant_df$isotype))){stop("isotype.to.plot input not found in dataframe. Please check if the isotype is spelled correctly")}
variant_df$match_ex_crit <- variant_df$isotype %in% isotypes.to.plot
variant_df <- subset(variant_df, subset = match_ex_crit == TRUE)
variant_df$match_ex_crit = NULL
variant_df$original_clonotypes = variant_df$clonotype_id
variant_df$clonotype_id = sprintf("%05d", variant_df$clonotype_id)
variant_df$clonotype_id = paste0("old", variant_df$clonotype_id)
rank_raw <- as.data.frame(variant_df %>% dplyr::group_by(clonotype_id) %>% dplyr::summarise(n = sum(n)))
rank_raw <- rank_raw[order(rank_raw$n, rank_raw$clonotype_id, decreasing = TRUE),]
rank_raw$new_clonotype <- 1:length(rank_raw$clonotype_id)
rank_raw$new_clonotype = sprintf("%05d", rank_raw$new_clonotype)
rank_raw$new_clonotype = paste0("new", rank_raw$new_clonotype)
for(z in 1:nrow(rank_raw)){
variant_df$clonotype_id <- gsub(rank_raw$clonotype_id[z], rank_raw$new_clonotype[z], variant_df$clonotype_id) #For VDJ
}
variant_df$clonotype_id <- readr::parse_number(variant_df$clonotype_id)
variant_df <- variant_df[order(variant_df$clonotype_id),]
message("New ranking based only on selected isotypes: ")
message(paste0(unique(variant_df$clonotype_id),", "))
}
variant_df$variant = 1:length(variant_df$clonotype_id)
variant_df$variant = as.character(variant_df$variant)
variant_df_list[[i]] = variant_df
#Delete clones not of interest
if (species == "Human"){
output_plot[[i]] <- ggplot2::ggplot(variant_df_list[[i]], ggplot2::aes(x = clonotype_id, y = n, fill = isotype, color = variant)) +
ggplot2::geom_bar(stat="identity", width=0.6, color="white") +
ggplot2::theme_bw() +
ggplot2::scale_fill_manual("Isotype", values = c("IGHG1" = "green","IGHG2" = "green3", "IGHG3"="green4", "IGHG4"="darkgreen", "IGHM" = "black", "IGHA1" = "red", "IGHA2"= "red4", "IGHD"="blue", "IGHE"="purple", "Unknown"="gray")) +
ggplot2::theme_classic() +
ggplot2::ggtitle(paste0(i)) +
ggplot2::scale_y_continuous(expand = c(0,0)) +
ggplot2::scale_x_continuous(expand = c(0,0.5)) +
ggplot2::labs(title = sample.names[[i]], x = "Clonal rank", y = "Number of cells") +
ggplot2::theme(text = ggplot2::element_text(size = 30),
plot.title = ggplot2::element_text(hjust = 0.5))
}
if (species == "Mouse"){
output_plot[[i]] <- ggplot2::ggplot(variant_df_list[[i]], ggplot2::aes(x = clonotype_id, y = n, fill = isotype, color = variant)) +
ggplot2::geom_bar(stat="identity", width=0.6, color="white") +
ggplot2::theme_bw() +
ggplot2::scale_fill_manual("Isotype", values = c("IGHG1" = "lightgreen", "IGHG2A"="green", "IGHG2B" = "green3", "IGHG2C"="green4", "IGHG3"="darkgreen", "IGHM" = "black", "IGHA" = "#fb6a4a", "IGHD"="#1874CD", "IGHE"="#B452CD", "Unknown"="gray")) +
ggplot2::theme_classic() +
ggplot2::ggtitle(paste0(i)) +
ggplot2::scale_y_continuous(expand = c(0,0)) +
ggplot2::scale_x_continuous(expand = c(0,0.5)) +
ggplot2::labs(title = sample.names[[i]], x = "Clonal rank", y = "Number of cells") +
ggplot2::theme(text = ggplot2::element_text(size = 30),
plot.title = ggplot2::element_text(hjust = 0.5))
}
}
clones_per_isotype_all = variant_df_list
names(clones_per_isotype_all) <- sample.names
names(output_plot) <- sample.names
return(list(output_plot,clones_per_isotype_all))
}
}
}else if(celltype == "Tcells"){ ####START T cells
#check that dataframe is the input
if(!inherits(VDJ.matrix,"data.frame")){
stop("Please provide a VDJ matrix dataframe (VDJ_GEX_matrix_output[[1]])")
}
#set up for subsetting by group.by item
VDJ.matrix[,group.by] <- as.character(VDJ.matrix[,group.by])
sample.names <- unique(VDJ.matrix[,group.by])
sample.names[is.na(sample.names)] <- "NONE"
VDJ.matrix[is.na(VDJ.matrix[,group.by]),group.by] <- "NONE"
VDJ.matrix.all <- VDJ.matrix
if(clones<1 | clones>50){stop("Number of clones must be an integer value between 1 and 50")}
VDJ_per_clone_output_all <- list()
clones_per_isotype <- list()
clones_per_isotype_all <-list()
output_plot <- list()
for (i in 1:length(sample.names)){
#subset VDJ.matrix by group.by
VDJ.matrix <- subset(VDJ.matrix.all, VDJ.matrix.all[,group.by] == sample.names[i])
#get clonotype frequency table
clono_freq <- as.data.frame(table(VDJ.matrix$clonotype_id))
clono_freq <- clono_freq[order(clono_freq$Freq, decreasing = TRUE),]
#get essential info from VDJ_GEX_matrix
curr_rep_iso <- VDJ.matrix[,c("barcode","clonotype_id", "VDJ_cdr3_aa", "VJ_cdr3_aa")]
if(color.by[1] != "isotype"){ #add color info in case its needed
curr_rep_iso$colors <- as.character(VDJ.matrix[,color.by]) #get as character
curr_rep_iso$colors[which(is.na(curr_rep_iso$colors))] <- "None" #replace any NAs which would mess up string counting later
if(inherits(VDJ.matrix[,color.by],"factor")){ #REORDER Depending on class in VDJ.matrix. This is an attempt at conserving existing factor levels such as in the seurat_clusters column
curr_rep_iso$colors <- ordered(as.factor(curr_rep_iso$colors),levels = c(levels(VDJ.matrix[,color.by]), "None")) #reorder
} else {
curr_rep_iso$colors <- ordered(as.factor(curr_rep_iso$colors),levels = c(as.character(unique(VDJ.matrix[,color.by])), "None"))
}
} else{
curr_rep_iso$colors <- "none"
}
curr_rep_iso$isotype <- "none"
#iterate over clones and get isotype info
clones_per_isotype <- list()
for (j in 1:clones){
curr_clone <- curr_rep_iso[which(curr_rep_iso$clonotype_id == clono_freq[j,1]),]
if(nrow(curr_clone) > 0){
color_cur_clone <- unique(curr_clone$colors)
n_color_cur_clone <- length(unique(curr_clone$colors))
clones_per_isotype[[j]] <- data.frame("Counts"=rep(0, n_color_cur_clone), "Color"= color_cur_clone, "ClonalRank"=rep("", n_color_cur_clone), "clonotype_id" = rep(curr_clone$clonotype_id[1],n_color_cur_clone), "VDJ_cdr3_aa" = rep(curr_clone$VDJ_cdr3_aa[which(curr_clone$VDJ_cdr3_aa != "")][1],n_color_cur_clone), "VJ_cdr3_aa" = rep(curr_clone$VJ_cdr3_aa[which(curr_clone$VJ_cdr3_aa != "")][1],n_color_cur_clone), "barcode" = rep(paste0(curr_clone$barcode, collapse = ";"),n_color_cur_clone))
if(color.by[1] == "isotype"){
clones_per_isotype[[j]]$Color <- c("None")
clones_per_isotype[[j]]$Counts <- nrow(curr_clone)
} else {
for(k in 1:nrow(clones_per_isotype[[j]])){
clones_per_isotype[[j]]$Counts[k] <- stringr::str_count(paste0("/",paste0(curr_clone$colors,collapse = "/ /"),"/"), pattern = paste0("/",as.character(clones_per_isotype[[j]]$Color[k]), "/"))
}
}
clones_per_isotype[[j]]$ClonalRank <- j
} #end if(nrow(curr_clone) > 0)
}
clones_per_isotype_all[[i]] <- do.call("rbind", clones_per_isotype)
if(color.by[1] == "isotype"){
output_plot[[i]] <- ggplot2::ggplot(clones_per_isotype_all[[i]], ggplot2::aes(fill = Color, y=Counts, x=ClonalRank)) + ggplot2::geom_bar(stat="identity", width=0.6, color="black") + ggplot2::theme_bw() + ggplot2::scale_fill_manual(values = c("gray80")) + ggplot2::theme_classic() + ggplot2::ggtitle(paste0(i)) + ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5)) + ggplot2::scale_y_continuous(expand = c(0,0)) + ggplot2::scale_x_continuous(expand = c(0,0.5)) + ggplot2::labs(title = sample.names[[i]], x = "Clonal rank", y = "Number of cells")
} else {
output_plot[[i]] <- ggplot2::ggplot(clones_per_isotype_all[[i]], ggplot2::aes(fill = Color, y=Counts, x=ClonalRank)) + ggplot2::geom_bar(stat="identity", width=0.6, color="black") + ggplot2::theme_bw() + ggplot2::theme_classic() + ggplot2::ggtitle(paste0(i)) + ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5)) + ggplot2::scale_y_continuous(expand = c(0,0)) + ggplot2::scale_x_continuous(expand = c(0,0.5)) + ggplot2::labs(title = sample.names[[i]], x = "Clonal rank", y = "Number of cells", fill = color.by)+ ggplot2::scale_fill_manual(values = grDevices::rainbow(n = length(unique(clones_per_isotype_all[[i]]$Color))))
}
}
names(clones_per_isotype_all) <- sample.names
return(list(output_plot,clones_per_isotype_all))
}
}
Try the Platypus package in your browser
Any scripts or data that you put into this service are public.
Platypus documentation built on Oct. 18, 2024, 5:08 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Defines functions VDJ_clonal_expansion
Documented in VDJ_clonal_expansion
#' Flexible wrapper for clonal expansion barplots by isotype, GEX cluster etc.
#'
#'@description Clonal frequency plot displaying clonal expansion for either T and B cells with Platypus VDJ_build() output. For Platypus v2 plotting of B cell clonotype expansion and isotypes please refer to VDJ_isotypes_per_clone.
#' @param VDJ VDJ dataframe generated using the VDJ_build function.
#' @param celltype Character. Either "Tcells" or "Bcells". If set to Tcells bars will not be colored by default and the parameters treat_incomplete_cells, treat_incomplete_clones, subtypes and species are ignored. The color.by and group.by arguments work identically for both celltypes. If none provided it will detect this param from the celltype column.
#' @param clones numeric value indicating the number of clones to be considered for the clonal expansion plot. Default value is 50. For a standard plot more than 50 is discouraged. When showing only one - possibly rare - isotype via isotypes.to.plot it may be useful to set this number higher (e.g. 100-200)
#' @param subtypes Logical indicating whether to display isotype subtypes or not.
#' @param isotypes.to.plot Character vector. Defaults to "all". This can be set to any number of specific Isotypes, that are to be shown exclusively. For example, to show only clones containing IgG, input "IGHG". If only wanting to check clones with IgA and IgD input c("IGHA","IGHD"). Works equally if subtypes are set to TRUE. Is ignored if color.by is not set to "isotype"
#' @param species Character indicating whether the samples are from "Mouse" or "Human". Default is "Human".
#' @param treat.incomplete.clones Character indicating how to proceed with clonotypes lacking a VDJC (in other words, no cell within the clonotype has a VDJC). "exclude" removes these clonotypes from the analysis. "include" keeps these clonotypes in the analysis. In the plot they will appear has having an unknown isotype.
#' @param treat.incomplete.cells Character indicating how to proceed with cells assigned to a clonotype but missing a VDJC. "proportional" to fill in the VDJ isotype according to the proportions present in of clonotype (in case present proportions are not replicable in the total number of cells e.g. 1/3 in 10 cells, values are rounded to the next full integer and if the new counts exceed the total number of cells, 1 is subtracted from the isotype of highest frequency. If the number is below the number of cell, 1 is added to the isotype with lowest frequency to preserve diversity), "exclude" to exclude them from analysis and rank clonotypes only by the number of cells with a heavy chain. This ranking may deviate from the frequency column in the clonotype table. CAVE: if treat_incomplete_cells is set to "exclude", clonotypes lacking a VDJC entierly will be removed from the analysis. This results in a similar but not identical output as when treat_incomplete_clones is set to true. The two parameters are thereby non-redundant.
#' @param group.by Character. Defaults to "sample_id". Column name of VDJ to split VDJ by. For each unique entry in that column a plot will be generated. Therefore plots can be generated by sample_id, group_id or any other metadata item.To get plots for the whole repertoire set to "none"
#' @param color.by Character. Defaults to "isotype". If set to "isotype" bars are colored by the respective IgH chain or in grey for T cells. This can alternatively be set to any column name of the VDJ. This allows coloring clones by their V_gene usage or by GEX clusters
#' @param variant.plot Logical indicating whether to plot the output showing the variants or not.
#' @param text.size Numeric value indicating the size of the text in the plot. Default is 12.
#' @return Returns a nested list. out[[1]] are plots out[[2]] are raw datatables containing also barcode and CDR3 information
#' @export
#' @examples
#'clonal_out <- VDJ_clonal_expansion(VDJ = Platypus::small_vdj,
#' celltype = "Bcells", clones = 30,subtypes = FALSE, species = "Mouse"
#' ,treat.incomplete.clones = "exclude"
#' ,treat.incomplete.cells = "proportional")
#'clonal_out[[1]] #list of plots
#'clonal_out[[2]] #list of source dataframes
VDJ_clonal_expansion <- function(VDJ,
celltype,
clones,
subtypes,
isotypes.to.plot,
species,
treat.incomplete.clones,
treat.incomplete.cells,
group.by,
color.by,
variant.plot,
text.size){
#Adding compatibility with input naming scheme
VDJ.matrix <- VDJ
VDJ.matrix[is.na(VDJ.matrix)] <- 'unknown'
VDJ <- NULL
ClonalRank <- NULL
Counts <- NULL
sum_counts <- NULL
Isotype <- NULL
Color <- NULL
pasted_variants <- NULL
isotype <- NULL
colors <- NULL
match_ex_crit <- NULL
clonotype_id <- NULL
variant <- NULL
n <- NULL
#add a color scheme similar to Seurat's
ggplotColours <- function(n, h = c(0, 360) + 15){
if ((diff(h) %% 360) < 1) h[2] <- h[2] - 360/n
grDevices::hcl(h = (seq(h[1], h[2], length = n)), c = 100, l = 65)
}
if(missing(clones)) clones <- 50
if(missing(subtypes)) subtypes <- FALSE
if(missing(isotypes.to.plot)) isotypes.to.plot <- "all"
if(missing(species)) species <- "Human"
if(missing(treat.incomplete.cells)) treat.incomplete.cells <- "proportional"
if(missing(treat.incomplete.clones)) treat.incomplete.clones <- "exclude"
if(missing(text.size)) text.size <- 12
if(!treat.incomplete.cells %in% c("exclude", "proportional")){
stop("Please set treat.incomplete.cells to either 'proportional' or 'exclude'. 'Proportional' will assign cells of a clonotype missing a VDJ chain proportionally to the isotypes present in that clone (Default). 'exclude' will remove all cells missing a VDJ chain and thereby also alter clonotype frequencies")
}
if(!treat.incomplete.clones %in% c("exclude", "include")){
stop("Please set treat.incomplete.clones to either 'include' or 'exclude'. 'include' will show also clones which of which no cell has a VDJ chain. 'exclude' will remove such clones (default)")
}
platypus.version <- "v3"
if(missing(celltype)){
if(stringr::str_detect(VDJ.matrix$celltype[1], "B")){celltype = "Bcells"
} else if(stringr::str_detect(VDJ.matrix$celltype[1], "T")){celltype = "Tcells"
} else {stop("No celltype found in celltype column. celltype column must contain either 'B cells' or 'T cells'")}
}
if(missing(group.by)){
group.by <- "sample_id"}
if(missing(variant.plot)){
variant.plot <- FALSE}
if(group.by == "none"){
group.by <- "ungroup"
VDJ.matrix$ungroup <- 1
} else if(!group.by %in% names(VDJ.matrix) & group.by != "sample_id" & group.by == "none"){
stop("Please provide a valid column name of the VDJ.matrix in group.by")
}
if(missing(color.by)){
color.by <- "isotype"}
if(!color.by %in% names(VDJ.matrix) & color.by != "isotype"){
stop("Please provide a valid column name of the VDJ.matrix in color.by, or set to 'isotype' for default (also for Tcells)")
} else if(color.by %in% names(VDJ.matrix) & color.by != "isotype"){
unique_colors <- ggplotColours(n = length(unique(VDJ.matrix[,color.by])))
names(unique_colors) <- unique(VDJ.matrix[,color.by])
}
if(celltype == "Bcells"){ ####START Bcells
#check that dataframe is the input
if(!inherits(VDJ.matrix,"data.frame")){
stop("Please provide a VDJ dataframe. (VDJ_GEX_matrix_out[[1]]")
}
#set up for subsetting by group.by item
VDJ.matrix[,group.by] <- as.character(VDJ.matrix[,group.by])
sample.names <- unique(VDJ.matrix[,group.by])
sample.names[is.na(sample.names)] <- "NONE"
VDJ.matrix[is.na(VDJ.matrix[,group.by]),group.by] <- "NONE"
VDJ.matrix.all <- VDJ.matrix
#Disabled due to the possibility of having low abundance of isotypes of interest. If filtered by isotype, 50 clones may not be enought to get a decent plot.
#if(clones<1 | clones > 50){stop("Number of clones must be an integer value between 1 and 50")}
VDJ_per_clone_output_all <- list()
clones_per_isotype <- list()
clones_per_isotype_all <-list()
output_plot <- list()
variant_df_list <- list()
if(subtypes == FALSE | color.by != "isotype"){ #if colored by something else we don't care about subtypes...
if(variant.plot == FALSE){
for (i in 1:length(sample.names)){
#subset VDJ.matrix by group.by
VDJ.matrix <- subset(VDJ.matrix.all, VDJ.matrix.all[,group.by] == sample.names[i])
#get clonotype frequency table
clono_freq <- as.data.frame(table(VDJ.matrix$clonotype_id))
clono_freq <- clono_freq[order(clono_freq$Freq, decreasing = TRUE),]
#get essential info from VDJ_GEX_matrix
curr_rep_iso <- VDJ.matrix[,c("barcode","clonotype_id", "VDJ_cgene", "VDJ_cdr3_aa", "VJ_cdr3_aa")]
names(curr_rep_iso)[3] <- "isotype"
if(color.by != "isotype"){ #add color info in case its needed
curr_rep_iso$colors <- as.character(VDJ.matrix[,color.by]) #get as character
curr_rep_iso$colors[which(is.na(curr_rep_iso$colors))] <- "None" #replace any NAs which would mess up string counting later
if(inherits(VDJ.matrix[,color.by],"factor")){ #REORDER Depending on class in VDJ.matrix. This is an attempt at conserving existing factor levels such as in the seurat_clusters column
curr_rep_iso$colors <- ordered(as.factor(curr_rep_iso$colors),levels = c(levels(VDJ.matrix[,color.by]), "None")) #reorder
} else {
curr_rep_iso$colors <- ordered(as.factor(curr_rep_iso$colors),levels = c(as.character(unique(VDJ.matrix[,color.by])), "None"))
}
}
if(treat.incomplete.clones == "exclude"){
clones_to_del <-c()
for(k in 1:nrow(clono_freq)){
if(all(curr_rep_iso[which(curr_rep_iso$clonotype_id == clono_freq[k,1]),"isotype"] == "")){ #detect if no VDJC is available in this clonotype
clones_to_del <- append(clones_to_del, k)
}
}
if(length(clones_to_del > 0)){
clono_freq <- clono_freq[-clones_to_del,]} #remove these clones entirely
clono_freq <- clono_freq[order(clono_freq$Freq, decreasing = T),] #reorder, to make sure
}
#iterate over clones and get isotype info
clones_per_isotype <- list()
for (j in 1:clones){
curr_clone <- curr_rep_iso[which(curr_rep_iso$clonotype_id == clono_freq[j,1]),]
if(nrow(curr_clone) > 0){
if(color.by == "isotype"){
#str_split and discard potentially second chain to avoid overcounting in the case of a cell having 2 HC with 2 different isotypes
curr_clone$isotype <- stringr::str_split(curr_clone$isotype, ";", simplify = TRUE)[,1]
if(treat.incomplete.cells == "proportional" & stringr::str_detect(paste0(curr_clone$isotype,collapse = ";"), pattern = "IG")==TRUE){ #check that there is at least one IG entry
props <- table(curr_clone$isotype[which(stringr::str_detect(curr_clone$isotype, pattern = "IG")==TRUE)]) #getting proportions
n_total <- nrow(curr_clone) #getting total number of cells in clonotype
props <- round(props / sum(props) * n_total,0) #calculating new number of each isotype to match proportions
if(sum(props) > n_total){props[which.max(props)] <- props[which.max(props)] - 1
} else if(sum(props) < n_total){props[which.min(props)] <- props[which.min(props)] + 1} #catching rounding derived errors.
curr_clone$isotype <- rep.int(names(props), props) #new isotype column with the new number of isotypes. ! these are not in the original order !
} else if (treat.incomplete.cells == "proportional" & stringr::str_detect(paste0(curr_clone$isotype,collapse = ";"), pattern = "IG")==FALSE){ #if no entry is present
curr_clone$isotype <- "None"
}
clones_per_isotype[[j]] <- data.frame("Counts"=rep(0, 6), "Color"=rep("", 6), "Isotype"=rep("", 6), "ClonalRank"=rep("", 6), "clonotype_id" = rep(curr_clone$clonotype_id[1],6), "VDJ_cdr3_aa" = rep(curr_clone$VDJ_cdr3_aa[which(curr_clone$VDJ_cdr3_aa != "")][1],6), "VJ_cdr3_aa" = rep(curr_clone$VJ_cdr3_aa[which(curr_clone$VJ_cdr3_aa != "")][1],6), "barcode" = rep(paste0(curr_clone$barcode, collapse = ";"),6)) #to maintain clonotype information
clones_per_isotype[[j]]$Counts[1] <- sum(stringr::str_count(curr_clone$isotype, "IGHG"))
clones_per_isotype[[j]]$Counts[2] <- sum(stringr::str_count(curr_clone$isotype, "IGHM"))
clones_per_isotype[[j]]$Counts[3] <- sum(stringr::str_count(curr_clone$isotype, "IGHA"))
clones_per_isotype[[j]]$Counts[4] <- sum(stringr::str_count(curr_clone$isotype, "IGHD"))
clones_per_isotype[[j]]$Counts[5] <- sum(stringr::str_count(curr_clone$isotype, "IGHE"))
clones_per_isotype[[j]]$Counts[6] <- sum(stringr::str_count(curr_clone$isotype, "None"))
if(color.by == "isotype"){
clones_per_isotype[[j]]$Color <- c("green4", "black", "red", "blue", "purple", "gray")
clones_per_isotype[[j]]$Isotype <- c("IGHG", "IGHM", "IGHA", "IGHD", "IGHE", "Unknown")
} else {
clones_per_isotype[[j]]$Isotype <- names(which.max(table(curr_clone$colors)))
}
clones_per_isotype[[j]]$ClonalRank <- j
} else if(color.by != "isotype"){
color_cur_clone <- unique(curr_rep_iso$colors)
n_color_cur_clone <- length(unique(curr_rep_iso$colors))
clones_per_isotype[[j]] <- data.frame("Counts"=rep(0, n_color_cur_clone), "Color"= color_cur_clone, "ClonalRank"=rep("", n_color_cur_clone), "clonotype_id" = rep(curr_clone$clonotype_id[1],n_color_cur_clone), "VDJ_cdr3_aa" = rep(curr_clone$VDJ_cdr3_aa[which(curr_clone$VDJ_cdr3_aa != "")][1],n_color_cur_clone), "VJ_cdr3_aa" = rep(curr_clone$VJ_cdr3_aa[which(curr_clone$VJ_cdr3_aa != "")][1],n_color_cur_clone), "barcode" = rep(paste0(curr_clone$barcode, collapse = ";"),n_color_cur_clone))
for(k in 1:nrow(clones_per_isotype[[j]])){
clones_per_isotype[[j]]$Counts[k] <- stringr::str_count(paste0("/",paste0(curr_clone$colors,collapse = "/ /"),"/"), pattern = paste0("/", as.character(clones_per_isotype[[j]]$Color[k]),"/"))
}
clones_per_isotype[[j]]$ClonalRank <- j
} # end not coloring by isotype but someting else
} #end if(nrow(curr_clone) > 0)
}
clones_per_isotype_all[[i]] <- do.call("rbind", clones_per_isotype)
if(treat.incomplete.cells == "exclude" & color.by == "isotype"){
rank_raw <- as.data.frame(clones_per_isotype_all[[i]] %>% dplyr::group_by(ClonalRank) %>% dplyr::summarise(sum_counts = sum(Counts)) %>% dplyr::arrange(dplyr::desc(sum_counts)) %>% dplyr::mutate(rank = 1:length(unique(ClonalRank))))
clones_per_isotype_all[[i]]$ClonalRank_2 <- 0
for(l in 1:nrow(rank_raw)){
clones_per_isotype_all[[i]]$ClonalRank_2[which(clones_per_isotype_all[[i]]$ClonalRank == rank_raw$ClonalRank[l])] <- rank_raw$rank[l]
}
clones_per_isotype_all[[i]]$ClonalRank <- clones_per_isotype_all[[i]]$ClonalRank_2
#reorder the dataframe the sum_counts order by decreasing sum, the clonal rank is there to keep clones that have the same count sum together as a group of rows
message("New ranking based only on present VDJ chains: ")
message(unique(clones_per_isotype_all[[i]]$clonotype_id))
}
#Delete clones not of interest
if(color.by == "isotype" & isotypes.to.plot[1] != "all"){
if(!any(isotypes.to.plot %in% unique(clones_per_isotype_all[[i]]$Isotype))){stop("isotype.to.plot input not found in dataframe. Please check if the isotype is spelled correctly")}
to_del <- c()
for(k in 1:length(unique(clones_per_isotype_all[[i]]$ClonalRank))){
if(sum(clones_per_isotype_all[[i]]$Counts[which(clones_per_isotype_all[[i]]$ClonalRank == unique(clones_per_isotype_all[[i]]$ClonalRank)[k] & clones_per_isotype_all[[i]]$Isotype %in% isotypes.to.plot)]) == 0){
to_del <- append(to_del, unique(clones_per_isotype_all[[i]]$ClonalRank)[k])
}
}
clones_per_isotype_all[[i]] <- subset(clones_per_isotype_all[[i]], !ClonalRank %in% to_del)
#redistribute clonal ranks
rank_raw <- as.data.frame(clones_per_isotype_all[[i]] %>% dplyr::group_by(ClonalRank) %>% dplyr::summarise(sum_counts = sum(Counts)) %>% dplyr::arrange(dplyr::desc(sum_counts)) %>% dplyr::mutate(rank = 1:length(unique(ClonalRank))))
clones_per_isotype_all[[i]]$ClonalRank_2 <- 0
for(l in 1:nrow(rank_raw)){
clones_per_isotype_all[[i]]$ClonalRank_2[which(clones_per_isotype_all[[i]]$ClonalRank == rank_raw$ClonalRank[l])] <- rank_raw$rank[l]
}
clones_per_isotype_all[[i]]$ClonalRank <- clones_per_isotype_all[[i]]$ClonalRank_2
message("New ranking based only on selected isotypes: ")
message(unique(clones_per_isotype_all[[i]]$clonotype_id))
}
if(color.by == "isotype"){
output_plot[[i]] <- ggplot2::ggplot(clones_per_isotype_all[[i]], ggplot2::aes(fill = Isotype, y=Counts, x=ClonalRank)) +
ggplot2::geom_bar(stat="identity", width=0.6, color="black") +
ggplot2::theme_bw() +
ggplot2::scale_fill_manual(values = ggplot2::scale_fill_manual(values = c("IGHG" = "green4", "IGHM" = "black", "IGHA" = "red3", "IGHD"="blue", "IGHE"="purple", "Unknown"="gray"))) +
ggplot2::theme_classic() +
ggplot2::ggtitle(paste0(i)) +
ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5), text = ggplot2::element_text(size = text.size)) +
ggplot2::scale_y_continuous(expand = c(0,0)) + ggplot2::scale_x_continuous(expand = c(0,0.5)) +
ggplot2::labs(title = sample.names[[i]], x = "Clonal rank", y = "Number of cells")
} else {
output_plot[[i]] <- ggplot2::ggplot(clones_per_isotype_all[[i]], ggplot2::aes(fill = Color, y=Counts, x=ClonalRank)) + ggplot2::geom_bar(stat="identity", width=0.6, color="black") + ggplot2::theme_bw() + ggplot2::theme_classic() + ggplot2::ggtitle(paste0(i)) + ggplot2::theme(text = ggplot2::element_text(size = text.size), plot.title = ggplot2::element_text(hjust = 0.5)) + ggplot2::scale_y_continuous(expand = c(0,0)) + ggplot2::scale_x_continuous(expand = c(0,0.5)) + ggplot2::labs(title = sample.names[[i]], x = "Clonal rank", y = "Number of cells", fill = color.by) + ggplot2::scale_fill_manual(values = unique_colors,labels=names(unique_colors))
}
}
names(clones_per_isotype_all) <- sample.names
return(list(output_plot,clones_per_isotype_all))
}
if(variant.plot ==TRUE){
for (i in 1:length(sample.names)){
message(paste0("Starting sample ", i, "/", length(sample.names)))
#subset VDJ.matrix by group.by
VDJ.matrix <- subset(VDJ.matrix.all, VDJ.matrix.all[,group.by] == sample.names[i])
#Rank clonotypes
s_id = VDJ.matrix$clonotype_id
s_id_num = readr::parse_number(s_id)
s_id_num = sprintf("%05d", s_id_num)
s_id_num = paste0("clonotype", s_id_num)
VDJ.matrix$clonotype_id <- s_id_num
s_id = NULL
s_id_num = NULL
VDJ.matrix$clonotype_id = paste0("old", VDJ.matrix$clonotype_id)
rank_raw <- as.data.frame(VDJ.matrix %>% dplyr::group_by(clonotype_id) %>% dplyr::summarise(n = n()))
rank_raw <- rank_raw[order(rank_raw$n, rank_raw$clonotype_id, decreasing = TRUE),]
#rank_raw <- rank_raw[order(rank_raw$n, decreasing = T),]
rank_raw$new_clonotype <- 1:length(rank_raw$clonotype_id)
rank_raw$new_clonotype = sprintf("%05d", rank_raw$new_clonotype)
rank_raw$new_clonotype = paste0("new", rank_raw$new_clonotype)
for(z in 1:nrow(rank_raw)){
VDJ.matrix$clonotype_id <- gsub(rank_raw$clonotype_id[z], rank_raw$new_clonotype[z], VDJ.matrix$clonotype_id) #For VDJ
}
VDJ.matrix$clonotype_id <- paste0("clonotype", sprintf("%05d", readr::parse_number(VDJ.matrix$clonotype_id)))
#get clonotype frequency table
clono_freq <- as.data.frame(table(VDJ.matrix$clonotype_id))
clono_freq <- clono_freq[order(clono_freq$Freq, decreasing = TRUE),]
is.trimmed = VDJ.matrix$VJ_sequence_nt_trimmed
#Define the variants
if (is.null(is.trimmed) == FALSE){
VDJ.matrix$pasted_variants = paste(VDJ.matrix$VDJ_sequence_nt_trimmed, VDJ.matrix$VJ_sequence_nt_trimmed, sep = ";")
message(paste0("Trimmed sequences found for sample ", sample.names[i]))
message("Variants are obtained as VDJ_sequence_nt_trimmed;VJ_sequence_nt_trimmed")
}
if (is.null(is.trimmed) == TRUE){
VDJ.matrix$pasted_variants = paste(VDJ.matrix$VDJ_sequence_nt_raw, VDJ.matrix$VJ_sequence_nt_raw, sep = ";")
message(paste0("Trimmed sequences not found for sample ", sample.names[i]))
message("Variants are obtained as VDJ_sequence_nt_raw;VJ_sequence_nt_raw")
}
#get essential info from VDJ_GEX_matrix
curr_rep_iso <- VDJ.matrix[,c("barcode","clonotype_id", "VDJ_cgene", "pasted_variants")]
names(curr_rep_iso)[3] <- "isotype"
#We only care about the main isotypes, not the subgroups
curr_rep_iso$isotype <- substr(curr_rep_iso$isotype,1,4)
if(color.by != "isotype"){ #add color info in case its needed
curr_rep_iso$colors <- as.character(VDJ.matrix[,color.by]) #get as character
curr_rep_iso$colors[which(is.na(curr_rep_iso$colors))] <- "None" #replace any NAs which would mess up string counting later
if(inherits(VDJ.matrix[,color.by],"factor")){ #REORDER Depending on class in VDJ.matrix. This is an attempt at conserving existing factor levels such as in the seurat_clusters column
curr_rep_iso$colors <- ordered(as.factor(curr_rep_iso$colors),levels = c(levels(VDJ.matrix[,color.by]), "None")) #reorder
} else {
curr_rep_iso$colors <- ordered(as.factor(curr_rep_iso$colors),levels = c(as.character(unique(VDJ.matrix[,color.by])), "None"))
}
}
if(treat.incomplete.clones == "exclude"){
clones_to_del <-c()
for(k in 1:nrow(clono_freq)){
if(all(curr_rep_iso[which(curr_rep_iso$clonotype_id == clono_freq[k,1]),"isotype"] == "")){ #detect if no VDJC is available in this clonotype
clones_to_del <- append(clones_to_del, k)
}
}
if(length(clones_to_del > 0)){
#Formatting numbers inside the function
s_id_num = clones_to_del
s_id_num = sprintf("%05d", s_id_num)
s_id_num = paste0("clonotype", s_id_num)
clones_to_del <- s_id_num
s_id = NULL
s_id_num = NULL
s_id = curr_rep_iso$clonotype_id
s_id_num = readr::parse_number(s_id)
s_id_num = sprintf("%05d", s_id_num)
s_id_num = paste0("clonotype", s_id_num)
curr_rep_iso$clonotype_id <- s_id_num
s_id = NULL
s_id_num = NULL
#Delete the clones that are incomplete
curr_rep_iso$match = curr_rep_iso$clonotype_id %in% clones_to_del
curr_rep_iso <- subset(curr_rep_iso, subset = match == FALSE)
curr_rep_iso$match = NULL
curr_rep_iso= curr_rep_iso[order(curr_rep_iso$clonotype_id),]
curr_rep_iso$clonotype_id <- paste0("old", curr_rep_iso$clonotype_id)
#Re rank the clonotypes
rank_raw <- as.data.frame(curr_rep_iso %>% dplyr::group_by(clonotype_id) %>% dplyr::summarise(n = n()))
rank_raw <- rank_raw[order(rank_raw$n, rank_raw$clonotype_id, decreasing = TRUE),]
rank_raw$new_clonotype <- 1:length(rank_raw$clonotype_id)
rank_raw$new_clonotype = paste0("newclonotype", sprintf("%05d", rank_raw$new_clonotype))
#Replace in the dataframe the old clonotypes by the new rank
for(z in 1:nrow(rank_raw)){
curr_rep_iso$clonotype_id <- gsub(rank_raw$clonotype_id[z], rank_raw$new_clonotype[z], curr_rep_iso$clonotype_id) #For VDJ
}
curr_rep_iso <- curr_rep_iso[order(curr_rep_iso$clonotype_id),]
}
}
#Formatting the clonotypes for the function
s_id = curr_rep_iso$clonotype_id
s_id_num = readr::parse_number(s_id)
s_id_num = sprintf("%05d", s_id_num)
s_id_num = paste0("clonotype", s_id_num)
curr_rep_iso$clonotype_id <- s_id_num
s_id = NULL
s_id_num = NULL
options(dplyr.summarise.inform = FALSE)
#Obtain the variant df depending on coloring
if(color.by == "isotype"){
curr_rep_iso %>% dplyr::group_by(pasted_variants, clonotype_id, isotype) %>% dplyr::summarise(n=n()) -> variant_df
}
if(color.by != "isotype"){
curr_rep_iso %>% dplyr::group_by(pasted_variants, clonotype_id, colors) %>% dplyr::summarise(n=n()) -> variant_df
}
variant_df= variant_df[order(variant_df$clonotype_id,variant_df$n),]
#Simplify the variant_df to keep only the number of clones desired
max_row = max(which(variant_df$clonotype_id == paste0("clonotype", sprintf("%05d", clones))))
variant_df = variant_df[1:max_row,]
s_id = variant_df$clonotype_id
s_id_num = readr::parse_number(s_id)
variant_df$clonotype_id <- s_id_num
s_id = NULL
s_id_num = NULL
if(treat.incomplete.cells == "proportional" & stringr::str_detect(paste0(variant_df$isotype,collapse = ";"), pattern = "IG")==TRUE){ #check that there is at least one IG entry
for (clone_number in 1:clones){
subset_clone_variant_with_isotype = variant_df[which(variant_df$clonotype_id == clone_number & stringr::str_detect(variant_df$isotype, pattern = "IG")==TRUE),]
subset_clone_variant = variant_df[which(variant_df$clonotype_id == clone_number),]
variants <- c()
if ((nrow(subset_clone_variant_with_isotype) < nrow(subset_clone_variant)) & (nrow(subset_clone_variant_with_isotype) >= 1)){
for (z in 1:nrow(subset_clone_variant_with_isotype)){
variants <- c(variants, rep(subset_clone_variant_with_isotype$isotype[z], subset_clone_variant_with_isotype$n[z]))
}
proportional_variants = c()
for (variantes in 1:length(which(variant_df$clonotype_id == clone_number & stringr::str_detect(variant_df$isotype, pattern = "IG")==FALSE))){
proportional_variants= c(proportional_variants, sample(variants, 1))
}
variant_df[which(variant_df$clonotype_id == clone_number & stringr::str_detect(variant_df$isotype, pattern = "IG")==FALSE),]$isotype = proportional_variants
} else if ((nrow(subset_clone_variant_with_isotype) < nrow(subset_clone_variant)) & (nrow(subset_clone_variant_with_isotype) == 0)){
proportional_variants = c()
proportional_variants = c(proportional_variants, rep("Unknown", nrow(subset_clone_variant)))
variant_df[which(variant_df$clonotype_id == clone_number & stringr::str_detect(variant_df$isotype, pattern = "IG")==FALSE),]$isotype = proportional_variants
}
}
}
if(treat.incomplete.cells == "exclude" & color.by == "isotype"){
variant_df$match_ex_crit <- variant_df$isotype == ""
variant_df <- subset(variant_df, subset = match_ex_crit == FALSE)
variant_df$match_ex_crit = NULL
variant_df$original_clonotypes = variant_df$clonotype_id
variant_df$clonotype_id = sprintf("%05d", variant_df$clonotype_id)
variant_df$clonotype_id = paste0("old", variant_df$clonotype_id)
rank_raw <- as.data.frame(variant_df %>% dplyr::group_by(clonotype_id) %>% dplyr::summarise(n = sum(n)))
rank_raw <- rank_raw[order(rank_raw$n, rank_raw$clonotype_id, decreasing = TRUE),]
rank_raw$new_clonotype <- 1:length(rank_raw$clonotype_id)
rank_raw$new_clonotype = sprintf("%05d", rank_raw$new_clonotype)
rank_raw$new_clonotype = paste0("new", rank_raw$new_clonotype)
for(z in 1:nrow(rank_raw)){
variant_df$clonotype_id <- gsub(rank_raw$clonotype_id[z], rank_raw$new_clonotype[z], variant_df$clonotype_id) #For VDJ
}
variant_df$clonotype_id <- readr::parse_number(variant_df$clonotype_id)
variant_df <- variant_df[order(variant_df$clonotype_id),]
}
#Delete clones not of interest
if(color.by == "isotype" & isotypes.to.plot[1] != "all"){
if(!any(isotypes.to.plot %in% unique(variant_df$isotype))){stop("isotype.to.plot input not found in dataframe. Please check if the isotype is spelled correctly")}
variant_df$match_ex_crit <- variant_df$isotype %in% isotypes.to.plot
variant_df <- subset(variant_df, subset = match_ex_crit == TRUE)
variant_df$match_ex_crit = NULL
variant_df$original_clonotypes = variant_df$clonotype_id
variant_df$clonotype_id = sprintf("%05d", variant_df$clonotype_id)
variant_df$clonotype_id = paste0("old", variant_df$clonotype_id)
rank_raw <- as.data.frame(variant_df %>% dplyr::group_by(clonotype_id) %>% dplyr::summarise(n = sum(n)))
rank_raw <- rank_raw[order(rank_raw$n, rank_raw$clonotype_id, decreasing = TRUE),]
rank_raw$new_clonotype <- 1:length(rank_raw$clonotype_id)
rank_raw$new_clonotype = sprintf("%05d", rank_raw$new_clonotype)
rank_raw$new_clonotype = paste0("new", rank_raw$new_clonotype)
for(z in 1:nrow(rank_raw)){
variant_df$clonotype_id <- gsub(rank_raw$clonotype_id[z], rank_raw$new_clonotype[z], variant_df$clonotype_id) #For VDJ
}
variant_df$clonotype_id <- readr::parse_number(variant_df$clonotype_id)
variant_df <- variant_df[order(variant_df$clonotype_id),]
message("New ranking based only on selected isotypes: ")
message(paste0(unique(variant_df$clonotype_id),", "))
}
variant_df$variant = 1:length(variant_df$clonotype_id)
variant_df$variant = as.character(variant_df$variant)
variant_df_list[[i]] = variant_df
if(color.by == "isotype"){
output_plot[[i]] <- ggplot2::ggplot(variant_df_list[[i]], ggplot2::aes(x = clonotype_id, y = n, fill = isotype, color = variant)) + ggplot2::geom_bar(stat="identity", width=0.6, color="white") + ggplot2::theme_bw() + ggplot2::scale_fill_manual("Isotype",values = c("IGHG" = "green4", "IGHM" = "black", "IGHA" = "red3", "IGHD"="blue", "IGHE"="purple", "Unknown"="gray")) + ggplot2::theme_classic() + ggplot2::ggtitle(paste0(i)) + ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5)) + ggplot2::scale_y_continuous(expand = c(0,0)) + ggplot2::scale_x_continuous(expand = c(0,0.5)) + ggplot2::labs(title = sample.names[[i]], x = "Clonal rank", y = "Number of cells")
} else {
output_plot[[i]] <- ggplot2::ggplot(variant_df_list[[i]], ggplot2::aes(x = clonotype_id, y = n, fill = colors, color = variant)) + ggplot2::geom_bar(stat="identity", width=0.6, color="white") + ggplot2::scale_fill_manual(values = grDevices::rainbow(n = length(unique(clones_per_isotype_all[[i]]$colors)))) + ggplot2::theme_classic() + ggplot2::ggtitle(paste0(i)) + ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5)) + ggplot2::scale_y_continuous(expand = c(0,0)) + ggplot2::scale_x_continuous(expand = c(0,0.5)) + ggplot2::labs(title = sample.names[[i]], x = "Clonal rank", y = "Number of cells")
}
}
clones_per_isotype_all = variant_df_list
names(clones_per_isotype_all) <- sample.names
names(output_plot) <- sample.names
return(list(output_plot,clones_per_isotype_all))
}
}
if(subtypes ==TRUE & color.by == "isotype"){
if(variant.plot == FALSE){
for (i in 1:length(sample.names)){
#subset VDJ.matrix by group.by
VDJ.matrix <- subset(VDJ.matrix.all, VDJ.matrix.all[,group.by] == sample.names[i])
#get clonotype frequency table
clono_freq <- as.data.frame(table(VDJ.matrix$clonotype_id))
clono_freq <- clono_freq[order(clono_freq$Freq, decreasing = TRUE),]
#get essential info from VDJ_GEX_matrix
curr_rep_iso <- VDJ.matrix[,c("barcode","clonotype_id", "VDJ_cgene", "VDJ_cdr3_aa", "VJ_cdr3_aa")]
curr_rep_iso$VDJ_cgene[is.null(curr_rep_iso$VDJ_cgene)] <- "none"
names(curr_rep_iso)[3] <- "isotype"
#no substring here, just splitting on a ; to exclude multiple isotypes of one clone
curr_rep_iso$isotype <- stringr::str_split(curr_rep_iso$isotype, ";", simplify = TRUE)[,1]
if(treat.incomplete.clones == "exclude"){
clones_to_del <-c()
for(k in 1:nrow(clono_freq)){
if(all(curr_rep_iso[which(curr_rep_iso$clonotype_id == clono_freq[k,1]),"isotype"] == "")){ #detect if no VDJC is available in this clonotype
clones_to_del <- append(clones_to_del, k)
}
}
if(length(clones_to_del > 0)){
clono_freq <- clono_freq[-clones_to_del,]} #remove these clones entirely
clono_freq <- clono_freq[order(clono_freq$Freq, decreasing = TRUE),] #reorder, to make sure
}
#iterate over clones and get isotype info
clones_per_isotype <- list()
j <- 1
for (j in 1:clones){
curr_clone <- curr_rep_iso[which(curr_rep_iso$clonotype_id == clono_freq[j,1]),]
if(nrow(curr_clone) > 0){
#str_split and discard potentially second chain to avoid overcounting in the case of a cell having 2 HC with 2 different isotypes
curr_clone$isotype <- stringr::str_split(curr_clone$isotype, ";", simplify = TRUE)[,1]
if(treat.incomplete.cells == "proportional" & stringr::str_detect(paste0(curr_clone$isotype,collapse = ";"), pattern = "IG")==TRUE){ #check that there is at least one IG entry
props <- table(curr_clone$isotype[which(stringr::str_detect(curr_clone$isotype, pattern = "IG")==TRUE)]) #getting proportions
n_total <- nrow(curr_clone) #getting total number of cells in clonotype
props <- round(props / sum(props) * n_total,0) #calculating new number of each isotype to match proportions
if(sum(props) > n_total){props[which.max(props)] <- props[which.max(props)] - 1
} else if(sum(props) < n_total){props[which.min(props)] <- props[which.min(props)] + 1} #catching rounding derived errors.
curr_clone$isotype <- rep.int(names(props), props) #new isotype column with the new number of isotypes. ! these are not in the original order !
} else if (treat.incomplete.cells == "proportional" & stringr::str_detect(paste0(curr_clone$isotype,collapse = ";"), pattern = "IG")==FALSE){ #if no entry is present
curr_clone$isotype <- "None"
}
clones_per_isotype[[j]] <- data.frame("Counts"=rep(0, 14), "Color"=rep("", 14), "Isotype"=rep("", 14), "ClonalRank"=rep("", 14), "clonotype_id" = rep(curr_clone$clonotype_id[1],14), "VDJ_cdr3_aa" = rep(curr_clone$VDJ_cdr3_aa[which(curr_clone$VDJ_cdr3_aa != "")][1],14), "VJ_cdr3_aa" = rep(curr_clone$VJ_cdr3_aa[which(curr_clone$VJ_cdr3_aa != "")][1],14), "barcode" = rep(paste0(curr_clone$barcode, collapse = ";"),14)) #to maintain clonotype information
clones_per_isotype[[j]]$Counts[1] <- sum(stringr::str_count(curr_clone$isotype, "IGHG1"))
clones_per_isotype[[j]]$Counts[2] <- sum(stringr::str_count(curr_clone$isotype, "IGHG2"))
clones_per_isotype[[j]]$Counts[3] <- sum(stringr::str_count(curr_clone$isotype, "IGHG2A"))
clones_per_isotype[[j]]$Counts[4] <- sum(stringr::str_count(curr_clone$isotype, "IGHG2B"))
clones_per_isotype[[j]]$Counts[5] <- sum(stringr::str_count(curr_clone$isotype, "IGHG2C"))
clones_per_isotype[[j]]$Counts[6] <- sum(stringr::str_count(curr_clone$isotype, "IGHG3"))
clones_per_isotype[[j]]$Counts[7] <- sum(stringr::str_count(curr_clone$isotype, "IGHG4"))
clones_per_isotype[[j]]$Counts[8] <- sum(stringr::str_count(curr_clone$isotype, "IGHM"))
clones_per_isotype[[j]]$Counts[9] <- sum(stringr::str_count(curr_clone$isotype, "IGHA"))
clones_per_isotype[[j]]$Counts[10] <- sum(stringr::str_count(curr_clone$isotype, "IGHA1"))
clones_per_isotype[[j]]$Counts[11] <- sum(stringr::str_count(curr_clone$isotype, "IGHA2"))
clones_per_isotype[[j]]$Counts[12] <- sum(stringr::str_count(curr_clone$isotype, "IGHD"))
clones_per_isotype[[j]]$Counts[13] <- sum(stringr::str_count(curr_clone$isotype, "IGHE"))
clones_per_isotype[[j]]$Counts[14] <- sum(stringr::str_count(curr_clone$isotype, "None"))
clones_per_isotype[[j]]$ClonalRank <- j
if(color.by == "isotype"){
clones_per_isotype[[j]]$Isotype <- c("IGHG1", "IGHG2", "IGHG2a", "IGHG2b", "IGHG2c", "IGHG3", "IGHG4", "IGHM", "IGHA", "IGHA1", "IGHA2", "IGHD", "IGHE", "Unknown")
if (species == "Human"){
clones_per_isotype[[j]] <- clones_per_isotype[[j]][-c(3, 4, 5, 9), ]
}
if (species == "Mouse"){
clones_per_isotype[[j]] <- clones_per_isotype[[j]][-c(2, 7, 10, 11), ]
}
} else {
clones_per_isotype[[j]]$Isotype <- names(which.max(table(curr_clone$colors)))
}
} #end if(nrow(curr_clone) > 0)
}
clones_per_isotype_all[[i]] <- do.call("rbind",clones_per_isotype)
if(treat.incomplete.cells == "exclude"){
rank_raw <- as.data.frame(clones_per_isotype_all[[i]] %>% dplyr::group_by(ClonalRank) %>% dplyr::summarise(sum_counts = sum(Counts)) %>% dplyr::arrange(dplyr::desc(sum_counts)) %>% dplyr::mutate(rank = 1:length(unique(ClonalRank))))
clones_per_isotype_all[[i]]$ClonalRank_2 <- 0
for(l in 1:nrow(rank_raw)){
clones_per_isotype_all[[i]]$ClonalRank_2[which(clones_per_isotype_all[[i]]$ClonalRank == rank_raw$ClonalRank[l])] <- rank_raw$rank[l]
}
clones_per_isotype_all[[i]]$ClonalRank <- clones_per_isotype_all[[i]]$ClonalRank_2
#reorder the dataframe the sum_counts order by decreasing sum, the clonal rank is there to keep clones that have the same count sum together as a group of rows
message("New ranking based only on present HC chains: ")
message(unique(clones_per_isotype_all[[i]]$clonotype_id))
}
#Delete clones not of interest
if(color.by == "isotype" & isotypes.to.plot[1] != "all"){
if(!any(isotypes.to.plot %in% unique(clones_per_isotype_all[[i]]$Isotype))){stop("isotype.to.plot input not found in dataframe. Please check if the isotype is spelled correctly")}
to_del <- c()
for(k in 1:length(unique(clones_per_isotype_all[[i]]$ClonalRank))){
if(sum(clones_per_isotype_all[[i]]$Counts[which(clones_per_isotype_all[[i]]$ClonalRank == unique(clones_per_isotype_all[[i]]$ClonalRank)[k] & clones_per_isotype_all[[i]]$Isotype %in% isotypes.to.plot)]) == 0){
to_del <- append(to_del, unique(clones_per_isotype_all[[i]]$ClonalRank)[k])
}
}
clones_per_isotype_all[[i]] <- subset(clones_per_isotype_all[[i]], !ClonalRank %in% to_del)
#redistribute clonal ranks
rank_raw <- as.data.frame(clones_per_isotype_all[[i]] %>% dplyr::group_by(ClonalRank) %>% dplyr::summarise(sum_counts = sum(Counts)) %>% plyr::arrange(dplyr::desc(sum_counts)) %>% dplyr::mutate(rank = 1:length(unique(ClonalRank))))
clones_per_isotype_all[[i]]$ClonalRank_2 <- 0
for(l in 1:nrow(rank_raw)){
clones_per_isotype_all[[i]]$ClonalRank_2[which(clones_per_isotype_all[[i]]$ClonalRank == rank_raw$ClonalRank[l])] <- rank_raw$rank[l]
}
clones_per_isotype_all[[i]]$ClonalRank <- clones_per_isotype_all[[i]]$ClonalRank_2
message("New ranking based only on selected isotypes: ")
message(unique(clones_per_isotype_all[[i]]$clonotype_id))
}
if (species == "Human"){
output_plot[[i]] <- ggplot2::ggplot(clones_per_isotype_all[[i]], ggplot2::aes(fill = Isotype, y=Counts, x=ClonalRank)) + ggplot2::geom_bar(stat="identity", width=0.6, color="black") + ggplot2::theme_bw() + ggplot2::scale_fill_manual("Isotype", values = c("IGHG1" = "green","IGHG2" = "green3", "IGHG3"="green4", "IGHG4"="darkgreen", "IGHM" = "black", "IGHA1" = "red", "IGHA2"= "red4", "IGHD"="blue", "IGHE"="purple", "Unknown"="gray")) + ggplot2::theme_classic() + ggplot2::ggtitle(sample.names[i]) + ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5)) + ggplot2::scale_y_continuous(expand = c(0,0)) + ggplot2::ylab("Number of cells") + ggplot2::xlab("Clonal rank") + ggplot2::scale_x_continuous(expand = c(0,0.5))
}
if (species == "Mouse"){
output_plot[[i]] <- ggplot2::ggplot(clones_per_isotype_all[[i]], ggplot2::aes(fill = Isotype, y=Counts, x=ClonalRank)) + ggplot2::geom_bar(stat="identity", width=0.6, color="black") + ggplot2::theme_bw() + ggplot2::scale_fill_manual("Isotype", values = c("IGHG1" = "lightgreen", "IGHG2a"="green", "IGHG2b" = "green3", "IGHG2c"="green4", "IGHG3"="darkgreen", "IGHM" = "black", "IGHA" = "red", "IGHD"="blue", "IGHE"="purple", "Unknown"="gray")) + ggplot2::theme_classic() + ggplot2::ggtitle(sample.names[i]) + ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5)) + ggplot2::scale_y_continuous(expand = c(0,0)) + ggplot2::ylab("Number of cells") + ggplot2::xlab("Clonal rank") + ggplot2::scale_x_continuous(expand = c(0,0.5))
}
}
names(clones_per_isotype_all) <- sample.names
return(list(output_plot,clones_per_isotype_all))
}
if(variant.plot ==TRUE){
for (i in 1:length(sample.names)){
message(paste0("Starting sample ", i, "/", length(sample.names)))
#subset VDJ.matrix by group.by
VDJ.matrix <- subset(VDJ.matrix.all, VDJ.matrix.all[,group.by] == sample.names[i])
#Rank clonotypes
s_id = VDJ.matrix$clonotype_id
s_id_num = readr::parse_number(s_id)
s_id_num = sprintf("%05d", s_id_num)
s_id_num = paste0("clonotype", s_id_num)
VDJ.matrix$clonotype_id <- s_id_num
s_id = NULL
s_id_num = NULL
VDJ.matrix$clonotype_id = paste0("old", VDJ.matrix$clonotype_id)
rank_raw <- as.data.frame(VDJ.matrix %>% dplyr::group_by(clonotype_id) %>% dplyr::summarise(n = n()))
rank_raw <- rank_raw[order(rank_raw$n, rank_raw$clonotype_id, decreasing = T),]
rank_raw$new_clonotype <- 1:length(rank_raw$clonotype_id)
rank_raw$new_clonotype = sprintf("%05d", rank_raw$new_clonotype)
rank_raw$new_clonotype = paste0("new", rank_raw$new_clonotype)
for(z in 1:nrow(rank_raw)){
VDJ.matrix$clonotype_id <- gsub(rank_raw$clonotype_id[z], rank_raw$new_clonotype[z], VDJ.matrix$clonotype_id) #For VDJ
}
VDJ.matrix$clonotype_id <- paste0("clonotype", sprintf("%05d", readr::parse_number(VDJ.matrix$clonotype_id)))
#get clonotype frequency table
clono_freq <- as.data.frame(table(VDJ.matrix$clonotype_id))
clono_freq <- clono_freq[order(clono_freq$Freq, decreasing = TRUE),]
is.trimmed = VDJ.matrix$VJ_sequence_nt_trimmed
#Define the variants
if (is.null(is.trimmed) == FALSE){
VDJ.matrix$pasted_variants = paste(VDJ.matrix$VDJ_sequence_nt_trimmed, VDJ.matrix$VJ_sequence_nt_trimmed, sep = ";")
message(paste0("Trimmed sequences found for sample ", sample.names[i]))
message("Variants are obtained as VDJ_sequence_nt_trimmed;VJ_sequence_nt_trimmed")
}
if (is.null(is.trimmed) == TRUE){
VDJ.matrix$pasted_variants = paste(VDJ.matrix$VDJ_sequence_nt_raw, VDJ.matrix$VJ_sequence_nt_raw, sep = ";")
message(paste0("Trimmed sequences not found for sample ", sample.names[i]))
message("Variants are obtained as VDJ_sequence_nt_raw;VJ_sequence_nt_raw")
}
#get essential info from VDJ_GEX_matrix
#get essential info from VDJ_GEX_matrix
curr_rep_iso <- VDJ.matrix[,c("barcode","clonotype_id", "VDJ_cgene", "pasted_variants")]
names(curr_rep_iso)[3] <- "isotype"
#no substring here, just splitting on a ; to exclude multiple isotypes of one clone
curr_rep_iso$isotype <- stringr::str_split(curr_rep_iso$isotype, ";", simplify = TRUE)[,1]
if(treat.incomplete.clones == "exclude"){
clones_to_del <-c()
for(k in 1:nrow(clono_freq)){
if(all(curr_rep_iso[which(curr_rep_iso$clonotype_id == clono_freq[k,1]),"isotype"] == "")){ #detect if no VDJC is available in this clonotype
clones_to_del <- append(clones_to_del, k)
}
}
if(length(clones_to_del > 0)){
#Formatting numbers inside the function
s_id_num = clones_to_del
s_id_num = sprintf("%05d", s_id_num)
s_id_num = paste0("clonotype", s_id_num)
clones_to_del <- s_id_num
s_id = NULL
s_id_num = NULL
s_id = curr_rep_iso$clonotype_id
s_id_num = readr::parse_number(s_id)
s_id_num = sprintf("%05d", s_id_num)
s_id_num = paste0("clonotype", s_id_num)
curr_rep_iso$clonotype_id <- s_id_num
s_id = NULL
s_id_num = NULL
#Delete the clones that are incomplete
curr_rep_iso$match = curr_rep_iso$clonotype_id %in% clones_to_del
curr_rep_iso <- subset(curr_rep_iso, subset = match == FALSE)
curr_rep_iso$match = NULL
curr_rep_iso= curr_rep_iso[order(curr_rep_iso$clonotype_id),]
curr_rep_iso$clonotype_id <- paste0("old", curr_rep_iso$clonotype_id)
#Re rank the clonotypes
rank_raw <- as.data.frame(curr_rep_iso %>% dplyr::group_by(clonotype_id) %>% dplyr::summarise(n = n()))
rank_raw <- rank_raw[order(rank_raw$n, rank_raw$clonotype_id, decreasing = TRUE),]
rank_raw$new_clonotype <- 1:length(rank_raw$clonotype_id)
rank_raw$new_clonotype = paste0("newclonotype", sprintf("%05d", rank_raw$new_clonotype))
#Replace in the dataframe the old clonotypes by the new rank
for(z in 1:nrow(rank_raw)){
curr_rep_iso$clonotype_id <- gsub(rank_raw$clonotype_id[z], rank_raw$new_clonotype[z], curr_rep_iso$clonotype_id) #For VDJ
}
curr_rep_iso <- curr_rep_iso[order(curr_rep_iso$clonotype_id),]
}
}
#Formatting the clonotypes for the function
s_id = curr_rep_iso$clonotype_id
s_id_num = readr::parse_number(s_id)
s_id_num = sprintf("%05d", s_id_num)
s_id_num = paste0("clonotype", s_id_num)
curr_rep_iso$clonotype_id <- s_id_num
s_id = NULL
s_id_num = NULL
options(dplyr.summarise.inform = FALSE)
curr_rep_iso %>% dplyr::group_by(pasted_variants, clonotype_id, isotype) %>% dplyr::summarise(n=n()) -> variant_df
variant_df= variant_df[order(variant_df$clonotype_id,variant_df$n),]
#Simplify the variant_df to keep only the number of clones desired
max_row = max(which(variant_df$clonotype_id == paste0("clonotype", sprintf("%05d", clones))))
variant_df = variant_df[1:max_row,]
s_id = variant_df$clonotype_id
s_id_num = readr::parse_number(s_id)
variant_df$clonotype_id <- s_id_num
s_id = NULL
s_id_num = NULL
if(treat.incomplete.cells == "proportional" & stringr::str_detect(paste0(variant_df$isotype,collapse = ";"), pattern = "IG")==TRUE){ #check that there is at least one IG entry
for (clone_number in 1:clones){
subset_clone_variant_with_isotype = variant_df[which(variant_df$clonotype_id == clone_number & stringr::str_detect(variant_df$isotype, pattern = "IG")==TRUE),]
subset_clone_variant = variant_df[which(variant_df$clonotype_id == clone_number),]
variants <- c()
if ((nrow(subset_clone_variant_with_isotype) < nrow(subset_clone_variant)) & (nrow(subset_clone_variant_with_isotype) >= 1)){
for (z in 1:nrow(subset_clone_variant_with_isotype)){
variants <- c(variants, rep(subset_clone_variant_with_isotype$isotype[z], subset_clone_variant_with_isotype$n[z]))
}
proportional_variants = c()
for (variantes in 1:length(which(variant_df$clonotype_id == clone_number & stringr::str_detect(variant_df$isotype, pattern = "IG")==FALSE))){
proportional_variants= c(proportional_variants, sample(variants, 1))
}
variant_df[which(variant_df$clonotype_id == clone_number & stringr::str_detect(variant_df$isotype, pattern = "IG")==FALSE),]$isotype = proportional_variants
} else if ((nrow(subset_clone_variant_with_isotype) < nrow(subset_clone_variant)) & (nrow(subset_clone_variant_with_isotype) == 0)){
proportional_variants = c()
proportional_variants = c(proportional_variants, rep("Unknown", nrow(subset_clone_variant)))
variant_df[which(variant_df$clonotype_id == clone_number & stringr::str_detect(variant_df$isotype, pattern = "IG")==FALSE),]$isotype = proportional_variants
}
}
}
if(treat.incomplete.cells == "exclude" & color.by == "isotype"){
variant_df$match_ex_crit <- variant_df$isotype == ""
variant_df <- subset(variant_df, subset = match_ex_crit == FALSE)
variant_df$match_ex_crit = NULL
variant_df$original_clonotypes = variant_df$clonotype_id
variant_df$clonotype_id = sprintf("%05d", variant_df$clonotype_id)
variant_df$clonotype_id = paste0("old", variant_df$clonotype_id)
rank_raw <- as.data.frame(variant_df %>% dplyr::group_by(clonotype_id) %>% dplyr::summarise(n = sum(n)))
rank_raw <- rank_raw[order(rank_raw$n, rank_raw$clonotype_id, decreasing = TRUE),]
rank_raw$new_clonotype <- 1:length(rank_raw$clonotype_id)
rank_raw$new_clonotype = sprintf("%05d", rank_raw$new_clonotype)
rank_raw$new_clonotype = paste0("new", rank_raw$new_clonotype)
for(z in 1:nrow(rank_raw)){
variant_df$clonotype_id <- gsub(rank_raw$clonotype_id[z], rank_raw$new_clonotype[z], variant_df$clonotype_id) #For VDJ
}
variant_df$clonotype_id <- readr::parse_number(variant_df$clonotype_id)
variant_df <- variant_df[order(variant_df$clonotype_id),]
}
if(color.by == "isotype" & isotypes.to.plot[1] != "all"){
if(!any(isotypes.to.plot %in% unique(variant_df$isotype))){stop("isotype.to.plot input not found in dataframe. Please check if the isotype is spelled correctly")}
variant_df$match_ex_crit <- variant_df$isotype %in% isotypes.to.plot
variant_df <- subset(variant_df, subset = match_ex_crit == TRUE)
variant_df$match_ex_crit = NULL
variant_df$original_clonotypes = variant_df$clonotype_id
variant_df$clonotype_id = sprintf("%05d", variant_df$clonotype_id)
variant_df$clonotype_id = paste0("old", variant_df$clonotype_id)
rank_raw <- as.data.frame(variant_df %>% dplyr::group_by(clonotype_id) %>% dplyr::summarise(n = sum(n)))
rank_raw <- rank_raw[order(rank_raw$n, rank_raw$clonotype_id, decreasing = TRUE),]
rank_raw$new_clonotype <- 1:length(rank_raw$clonotype_id)
rank_raw$new_clonotype = sprintf("%05d", rank_raw$new_clonotype)
rank_raw$new_clonotype = paste0("new", rank_raw$new_clonotype)
for(z in 1:nrow(rank_raw)){
variant_df$clonotype_id <- gsub(rank_raw$clonotype_id[z], rank_raw$new_clonotype[z], variant_df$clonotype_id) #For VDJ
}
variant_df$clonotype_id <- readr::parse_number(variant_df$clonotype_id)
variant_df <- variant_df[order(variant_df$clonotype_id),]
message("New ranking based only on selected isotypes: ")
message(paste0(unique(variant_df$clonotype_id),", "))
}
variant_df$variant = 1:length(variant_df$clonotype_id)
variant_df$variant = as.character(variant_df$variant)
variant_df_list[[i]] = variant_df
#Delete clones not of interest
if (species == "Human"){
output_plot[[i]] <- ggplot2::ggplot(variant_df_list[[i]], ggplot2::aes(x = clonotype_id, y = n, fill = isotype, color = variant)) +
ggplot2::geom_bar(stat="identity", width=0.6, color="white") +
ggplot2::theme_bw() +
ggplot2::scale_fill_manual("Isotype", values = c("IGHG1" = "green","IGHG2" = "green3", "IGHG3"="green4", "IGHG4"="darkgreen", "IGHM" = "black", "IGHA1" = "red", "IGHA2"= "red4", "IGHD"="blue", "IGHE"="purple", "Unknown"="gray")) +
ggplot2::theme_classic() +
ggplot2::ggtitle(paste0(i)) +
ggplot2::scale_y_continuous(expand = c(0,0)) +
ggplot2::scale_x_continuous(expand = c(0,0.5)) +
ggplot2::labs(title = sample.names[[i]], x = "Clonal rank", y = "Number of cells") +
ggplot2::theme(text = ggplot2::element_text(size = 30),
plot.title = ggplot2::element_text(hjust = 0.5))
}
if (species == "Mouse"){
output_plot[[i]] <- ggplot2::ggplot(variant_df_list[[i]], ggplot2::aes(x = clonotype_id, y = n, fill = isotype, color = variant)) +
ggplot2::geom_bar(stat="identity", width=0.6, color="white") +
ggplot2::theme_bw() +
ggplot2::scale_fill_manual("Isotype", values = c("IGHG1" = "lightgreen", "IGHG2A"="green", "IGHG2B" = "green3", "IGHG2C"="green4", "IGHG3"="darkgreen", "IGHM" = "black", "IGHA" = "#fb6a4a", "IGHD"="#1874CD", "IGHE"="#B452CD", "Unknown"="gray")) +
ggplot2::theme_classic() +
ggplot2::ggtitle(paste0(i)) +
ggplot2::scale_y_continuous(expand = c(0,0)) +
ggplot2::scale_x_continuous(expand = c(0,0.5)) +
ggplot2::labs(title = sample.names[[i]], x = "Clonal rank", y = "Number of cells") +
ggplot2::theme(text = ggplot2::element_text(size = 30),
plot.title = ggplot2::element_text(hjust = 0.5))
}
}
clones_per_isotype_all = variant_df_list
names(clones_per_isotype_all) <- sample.names
names(output_plot) <- sample.names
return(list(output_plot,clones_per_isotype_all))
}
}
}else if(celltype == "Tcells"){ ####START T cells
#check that dataframe is the input
if(!inherits(VDJ.matrix,"data.frame")){
stop("Please provide a VDJ matrix dataframe (VDJ_GEX_matrix_output[[1]])")
}
#set up for subsetting by group.by item
VDJ.matrix[,group.by] <- as.character(VDJ.matrix[,group.by])
sample.names <- unique(VDJ.matrix[,group.by])
sample.names[is.na(sample.names)] <- "NONE"
VDJ.matrix[is.na(VDJ.matrix[,group.by]),group.by] <- "NONE"
VDJ.matrix.all <- VDJ.matrix
if(clones<1 | clones>50){stop("Number of clones must be an integer value between 1 and 50")}
VDJ_per_clone_output_all <- list()
clones_per_isotype <- list()
clones_per_isotype_all <-list()
output_plot <- list()
for (i in 1:length(sample.names)){
#subset VDJ.matrix by group.by
VDJ.matrix <- subset(VDJ.matrix.all, VDJ.matrix.all[,group.by] == sample.names[i])
#get clonotype frequency table
clono_freq <- as.data.frame(table(VDJ.matrix$clonotype_id))
clono_freq <- clono_freq[order(clono_freq$Freq, decreasing = TRUE),]
#get essential info from VDJ_GEX_matrix
curr_rep_iso <- VDJ.matrix[,c("barcode","clonotype_id", "VDJ_cdr3_aa", "VJ_cdr3_aa")]
if(color.by[1] != "isotype"){ #add color info in case its needed
curr_rep_iso$colors <- as.character(VDJ.matrix[,color.by]) #get as character
curr_rep_iso$colors[which(is.na(curr_rep_iso$colors))] <- "None" #replace any NAs which would mess up string counting later
if(inherits(VDJ.matrix[,color.by],"factor")){ #REORDER Depending on class in VDJ.matrix. This is an attempt at conserving existing factor levels such as in the seurat_clusters column
curr_rep_iso$colors <- ordered(as.factor(curr_rep_iso$colors),levels = c(levels(VDJ.matrix[,color.by]), "None")) #reorder
} else {
curr_rep_iso$colors <- ordered(as.factor(curr_rep_iso$colors),levels = c(as.character(unique(VDJ.matrix[,color.by])), "None"))
}
} else{
curr_rep_iso$colors <- "none"
}
curr_rep_iso$isotype <- "none"
#iterate over clones and get isotype info
clones_per_isotype <- list()
for (j in 1:clones){
curr_clone <- curr_rep_iso[which(curr_rep_iso$clonotype_id == clono_freq[j,1]),]
if(nrow(curr_clone) > 0){
color_cur_clone <- unique(curr_clone$colors)
n_color_cur_clone <- length(unique(curr_clone$colors))
clones_per_isotype[[j]] <- data.frame("Counts"=rep(0, n_color_cur_clone), "Color"= color_cur_clone, "ClonalRank"=rep("", n_color_cur_clone), "clonotype_id" = rep(curr_clone$clonotype_id[1],n_color_cur_clone), "VDJ_cdr3_aa" = rep(curr_clone$VDJ_cdr3_aa[which(curr_clone$VDJ_cdr3_aa != "")][1],n_color_cur_clone), "VJ_cdr3_aa" = rep(curr_clone$VJ_cdr3_aa[which(curr_clone$VJ_cdr3_aa != "")][1],n_color_cur_clone), "barcode" = rep(paste0(curr_clone$barcode, collapse = ";"),n_color_cur_clone))
if(color.by[1] == "isotype"){
clones_per_isotype[[j]]$Color <- c("None")
clones_per_isotype[[j]]$Counts <- nrow(curr_clone)
} else {
for(k in 1:nrow(clones_per_isotype[[j]])){
clones_per_isotype[[j]]$Counts[k] <- stringr::str_count(paste0("/",paste0(curr_clone$colors,collapse = "/ /"),"/"), pattern = paste0("/",as.character(clones_per_isotype[[j]]$Color[k]), "/"))
}
}
clones_per_isotype[[j]]$ClonalRank <- j
} #end if(nrow(curr_clone) > 0)
}
clones_per_isotype_all[[i]] <- do.call("rbind", clones_per_isotype)
if(color.by[1] == "isotype"){
output_plot[[i]] <- ggplot2::ggplot(clones_per_isotype_all[[i]], ggplot2::aes(fill = Color, y=Counts, x=ClonalRank)) + ggplot2::geom_bar(stat="identity", width=0.6, color="black") + ggplot2::theme_bw() + ggplot2::scale_fill_manual(values = c("gray80")) + ggplot2::theme_classic() + ggplot2::ggtitle(paste0(i)) + ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5)) + ggplot2::scale_y_continuous(expand = c(0,0)) + ggplot2::scale_x_continuous(expand = c(0,0.5)) + ggplot2::labs(title = sample.names[[i]], x = "Clonal rank", y = "Number of cells")
} else {
output_plot[[i]] <- ggplot2::ggplot(clones_per_isotype_all[[i]], ggplot2::aes(fill = Color, y=Counts, x=ClonalRank)) + ggplot2::geom_bar(stat="identity", width=0.6, color="black") + ggplot2::theme_bw() + ggplot2::theme_classic() + ggplot2::ggtitle(paste0(i)) + ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5)) + ggplot2::scale_y_continuous(expand = c(0,0)) + ggplot2::scale_x_continuous(expand = c(0,0.5)) + ggplot2::labs(title = sample.names[[i]], x = "Clonal rank", y = "Number of cells", fill = color.by)+ ggplot2::scale_fill_manual(values = grDevices::rainbow(n = length(unique(clones_per_isotype_all[[i]]$Color))))
}
}
names(clones_per_isotype_all) <- sample.names
return(list(output_plot,clones_per_isotype_all))
}
}
Try the Platypus package in your browser
Any scripts or data that you put into this service are public.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.