create_T2C_stats.R

#!/usr/bin/env Rscript
args = commandArgs(trailingOnly=TRUE)

require(plyr, quietly = TRUE)
require(data.table, quietly = TRUE)
require(grid, quietly = TRUE)
require(gridExtra, quietly = TRUE)

path = args[1]
outfile = args[2]

chr <- args[3]
start <- args[4] ## 137189494
end <- args[5]  ## 142385639
sample_name <- args[6]

reds.bed <- data.table::fread(path, header = FALSE, sep = "\t")
region_pairs <- subset(reds.bed, (V1 == chr & V2 >= start & V4 == chr & V6 <= end))
colnames(region_pairs) <- c("Chromosome_1", "Start_1", "End_1", "Chromosome_2", "Start_2", "End_2", "Strand", "Score" )

if(nrow(region_pairs) == 0){
  stop("ERROR: No Interaktions found in Target Region")
}

count <- ddply(region_pairs,.(Chromosome_1, Start_1, End_1, Chromosome_2, Start_2, End_2, Strand , Score),nrow)
mean_interactions <- round(mean(count[,9]))
median_interactions <- median(count[,9])
interactions <- nrow(unique(region_pairs))
stats <- data.table::data.table(c(interactions),
                                c(mean_interactions))
rownames(stats) <- c(sample_name)
colnames(stats) <- c("Number of all Interactions", "Average Number of Read-Pairs per Interaction")
pdf(outfile, height=10, width=12.5)
grid.table(stats)
dev.off()
graphics.off()