R/annoByGaps.R

In bioinfo16/IRFinder: Vector integrated region analysis package (IRFinder)

#' @title Annotate the vector integrated site by gap information from assembly.
#' 
#' @description This function uses gap data to search the feature of integrated regions. 
#'              User can get query sequence inserted in annotated gaps and 
#'              distribution graphes from annotated gaps coordinations such as centromere and telomere.
#'              Plus, user can do random distribution analysis by this function.
#' 
#' @usage annoByGaps(hits, randomSet = NULL, mapTool = 'blast', organism = 'hg19', interval = 5000, range = c(-20000, 20000),
#'                   gapType = c('centromere', 'telomere'), outpath = '~', dbPath = paste0(.libPaths()[1], '/IRFinder/extdata'))
#' 
#' @param hits a GR object. This object made from *makeInputs* function.
#' @param randomSet a string vector. Type path to load a random set. 
#'                  If this value is null, random distribution analysis is not executed.
#' @param mapTool a character. Function serve two types of file such as outputs from BLAST and BLAT.
#'                Default is 'blast'. If you want to use BLAT output, use 'blast'.
#' @param organism a single character. This function serves 3 versions of organisms such as hg19, hg38 (Human)
#'                 and galGal6 (Chicken). Default is 'hg19'.
#' @param interval an integer vector. This number means interval number for distribution analysis. Default is 5000.
#' @param range an integer array. It means the range for highlight region of this analysis. Default range is c(-20000, 20000).
#' @param gapType a character vector. User can select annotated gap types such as centromere, telomere and heterochromatin. 
#' @param outpath a string vector. Plots are saved in this path. Default value is R home directory.
#' @param dbPath a string vector. Directory path of database files.
#' 
#' @return Return a result list constituted by insertion table, distribution table and GRobject of Gap data.
#'
#'
#' @export

annoByGaps = function(hits, randomSet = NULL, mapTool = 'blast', organism = 'hg19', interval = 5000, 
                      range = c(-20000, 20000), gapType = c('centromere', 'telomere', 'heterochromatin'),
                      outpath = '~', dbPath = paste0(.libPaths()[1], '/IRFinder/extdata')){
  library(GenomicRanges); library(stringr); library(grDevices); library(regioneR)
  
  cat('---------- Annotation integrated sites : Annotated gaps ----------\n')
  cat(paste0('Start time : ', date(), '\n'))
  
  if(length(which(c('hg19', 'hg38', 'galGal6') %in% organism)) == 0){
    return(cat("You can use hg19/hg38/galGal6 data only. ( Input : ", paste(organism, collapse = ','), ")\n",
               '---------- Annotation process is halted. ----------\nFinish time : ', date(), '\n'))
  } else {}
  
  cat('---------- Loading a gap annotation table ----------\n')
  #### 01. Load a gap table
  tab_loc = paste0(dbPath, '/gap.txt.gz')
  tab_gz = gzfile(description = tab_loc, open = "r")
  suppressWarnings(dataTable <- read.delim(file = tab_gz, header = FALSE, stringsAsFactors = FALSE))
  dataTable = dataTable[,c(2,3,4,7,8)]
  colnames(dataTable) = c('chrom', 'start', 'end', 'size', 'type')
  close(tab_gz)
  
  dataTable = subset(dataTable, dataTable$type %in% gapType)
  dataTable[,2] = dataTable[,2]+1

  cat('Done.\n')
  cat('---------- Creating a GRanges object ----------\n')
  
  #### 02. Make GR object by a gap table
  gr_gaps = regioneR::toGRanges(dataTable)

  #### 03. Make interval GR objects of a gap table
  ranges = seq(from = range[1], to = range[2], interval)
  gr_gaps_dist = vector("list", length = (length(ranges)-1))
  for(x in 1:(length(ranges)-1)){
    tmp_start = gr_gaps@ranges@start + ranges[x]
    tmp_end = gr_gaps@ranges@start + ranges[x+1]-1
    gr_gaps_dist[[x]] = GRanges(seqnames = as.character(gr_gaps@seqnames),
                                 ranges = IRanges(start = tmp_start,
                                                  end = tmp_end),
                                 strand = '*')
  }

  #### 04. Make random GR object
  if(!is.null(randomSet)){
    tmp = read.delim(file = randomSet, header = TRUE, stringsAsFactors = FALSE)
    gr_random = regioneR::toGRanges(tmp[,c(2,3,3)])
  } else {
    cat("[WARN] Random distribution analysis will not be executed.\n")
  }
  
  cat('Done.\n')
  cat('---------- Annotating integrated regions  ----------\n')
  
  #### 05. Make gap information
  inside_gap = as.data.frame(findOverlaps(hits, gr_gaps, type = 'any',
                                           ignore.strand = TRUE),
                              stringsAsFactors = FALSE)
  a = as.data.frame(hits[inside_gap$queryHits,], stringsAsFactors = FALSE)
  b = dataTable[inside_gap$subjectHits,]
  inside_tab = cbind(a,b)
  
  if(!is.null(randomSet)){
    inside_gap_ran = as.data.frame(findOverlaps(gr_random, gr_gaps, type = 'any', ignore.strand = TRUE), 
                                    stringsAsFactors = FALSE)
    a = as.data.frame(gr_random[inside_gap_ran$queryHits,], stringsAsFactors = FALSE)
    b = dataTable[inside_gap_ran$subjectHits,]
    inside_ran_tab = cbind(a,b)
    
    dist_gap_ran = vector("list", length = length(ranges)-1)
    dist_gap_ran_tab = data.frame(stringsAsFactors = FALSE)
  } else {}
  
  dist_gap = vector("list", length = length(ranges)-1)
  dist_gap_tab = data.frame(stringsAsFactors = FALSE)

  for(x in 1:(length(ranges)-1)){
    dist_gap[[x]] = as.data.frame(findOverlaps(query = hits,
                                                subject = gr_gaps_dist[[x]],
                                                type = "any", ignore.strand = TRUE),
                                   stringsAsFactors = FALSE)
    a = as.data.frame(hits[dist_gap[[x]]$queryHits,], stringsAsFactors = FALSE)
    b = dataTable[dist_gap[[x]]$subjectHits,]
    tmp1 = cbind(a,b)
    dist_gap_tab = rbind(dist_gap_tab, tmp1)

    if(!is.null(randomSet)){
      dist_gap_ran[[x]] = as.data.frame(findOverlaps(query = gr_random,
                                                      subject = gr_gaps_dist[[x]],
                                                      type = "any", ignore.strand = TRUE),
                                         stringsAsFactors = FALSE)
      a = as.data.frame(gr_random[dist_gap_ran[[x]]$queryHits,], stringsAsFactors = FALSE)
      b = dataTable[dist_gap_ran[[x]]$subjectHits,]
      tmp1 = cbind(a,b)
      dist_gap_ran_tab = rbind(dist_gap_ran_tab, tmp1)
    } else {}
  }
  
  #### 06. Count the number of hit in each range
  count_hit_gap = vector("list", length = length(ranges)-1)

  if(!is.null(randomSet)){
    count_hit_gap_ran = vector("list", length = length(ranges)-1)
  } else {}
  
  for(x in 1:(length(ranges)-1)){
    count_hit_gap[[x]] = countOverlaps(query = hits, subject = gr_gaps_dist[[x]], type = "any", ignore.strand = TRUE)
    if(!is.null(randomSet)){
      count_hit_gap_ran[[x]] = countOverlaps(query = gr_random, subject = gr_gaps_dist[[x]], type = "any", ignore.strand = TRUE)
    } else {}
  }
  
  hits_gap = as.numeric(apply(as.data.frame(count_hit_gap, stringsAsFactors = FALSE), MARGIN = 2, sum))

  if(!is.null(randomSet)){
    hits_gap_ran = as.numeric(apply(as.data.frame(count_hit_gap_ran, stringsAsFactors = FALSE), MARGIN = 2, sum))
  } else {}
  
  #### 07. Drawing histograms
  ymax_g = max(hits_gap)
  
  mid = vector("numeric", length = (length(ranges)-1))
  for(x in 1:(length(ranges)-1)){
    mid[x] = (ranges[x] + ranges[x+1])/2
  }
  
  freq_sites_gap = rep(x = mid, times = hits_gap)

  if(!is.null(randomSet)){
    freq_sites_gap_ran = rep(x = mid, times = hits_gap_ran)
  } else {}
  
  cat('Done.\n')
  cat('---------- Drawing histograms ----------\n')
  
  png(filename = paste0(outpath, '/Distribution_gap_', organism, '.png'),
      width = 1200, height = 700)
  hist_frequency = hist(freq_sites_gap/1000, ylim = c(0, ceiling(ymax_g * 1.1)), breaks = ranges/1000,
                        ylab = '#Integration events', xlab = "kbs",
                        probability = FALSE, main = NULL, col = 'cornflowerblue',
                        axes = FALSE, cex.lab = 1.5)
  axis(side = 1, at = ranges/1000, cex.axis = 1.5)
  axis(side = 2, at = seq(0, ceiling(ymax_g * 1.1), 1), cex.axis = 1.5)
  text(0, ceiling(ymax_g * 1.1), labels = 'Gap', cex = 2, font = 2)
  arrows(0,0,0,ceiling(ymax_g * 1.1)*0.95, length = 0.15, code = 1, lwd = 3)
  arrows(min(ranges/1000), ceiling(ymax_g * 1.1)*0.95, range[1]/1000*0.025, ceiling(ymax_g * 1.1)*0.95, length = 0.1, code = 1)
  arrows(max(ranges/1000), ceiling(ymax_g * 1.1)*0.95, range[2]/1000*0.025, ceiling(ymax_g * 1.1)*0.95, length = 0.1, code = 1)
  text(range[2]/2000, ceiling(ymax_g * 1.1)*0.9, 'Upstream', cex = 1.5, col = 'black')
  text(-(range[2]/2000), ceiling(ymax_g * 1.1)*0.9, 'Downstream', cex = 1.5, col = 'black')
  dev.off()
  
  if(!is.null(randomSet)){
    count_sites_gap = plyr::count(freq_sites_gap)[,2]
    count_sites_gap_ran = plyr::count(freq_sites_gap_ran)[,2]
    
    count_data = rbind(count_sites_gap, count_sites_gap_ran)
    png(paste0(outpath, '/Random_distribution_gap_', organism, '.png'), width = 1200, height = 750)
    barplot(count_data, beside = TRUE, ylim = c(0, (max(count_data)+2)),
            main = "Random distribution (Gap)", xlab = 'Intervals (Kbs)', ylab = "#Integration events",
            col = c('Dodgerblue', 'skyblue'), names.arg = ranges[c(1:((length(ranges)-1)/2), ((length(ranges)+3)/2):length(ranges))])
    dev.off()
  } else {}
  
  if(!is.null(randomSet)){
    result_list = vector("list", length = 4)
    result_list[[1]] = inside_tab
    result_list[[2]] = dist_gap_tab
    result_list[[3]] = dist_gap_ran_tab
    result_list[[4]] = gr_gaps
    names(result_list) = c('Gap_inside_hits', 'Gap_exp_distribution', 'Gap_random_distribution', 'Gap_data',
                           'TSS_exp_distribution', 'TSS_random_distribution', 'TSS_data')
  } else {
    result_list = vector("list", length = 3)
    result_list[[1]] = inside_tab
    result_list[[2]] = dist_gap_tab
    result_list[[3]] = gr_gaps
    names(result_list) = c('Gap_inside_hits', 'Gap_exp_distribution', 'Gap_data', 'TSS_exp_distribution', 'TSS_data')
  }
  
  cat('---------- Annotation process is finished. ----------\n')
  cat(paste0('Finish time : ', date(), '\n'))
  
  return(result_list)
}