data-raw/telomere.R

In cancer-genomics/plasmasv: Methods for identification of tumor-specific structural variants in plasma

# gap_hg19.txt: gap track from the UCSC genome browser for the hg19 genome build. Downloaded April 10, 2020
# from https://genome.ucsc.edu/cgi-bin/hgTables with the following settings:
# clade: Mammal
# genome: Human
# assembly: Feb. 2009 (GRCh37/hg19)
# group: All Tables
# database: hg19
# table: gap

library(GenomicRanges)

gap.df <- read.table("/dcl01/scharpf1/data/dbruhm/delfi_followup/split-reads/pcawg_analysis/data/raw_data/gap_hg19.txt",
                     header = TRUE, stringsAsFactors = FALSE, comment.char = "")
chrs <- paste0("chr", c(seq(1, 22, 1), "X", "Y"))
gap.df <- subset(gap.df, gap.df$chrom %in% chrs)
gap.gr <- GRanges(seqnames = gap.df$chrom, ranges = IRanges(start = gap.df$chromStart, end = gap.df$chromEnd), type = gap.df$type)
telomere.gr <- gap.gr[gap.gr$type == "telomere"]
start(telomere.gr)[which(start(telomere) == 0)] <- 1

telomere <- telomere.gr

save(telomere, file = "/dcl01/scharpf1/data/dbruhm/delfi_followup/split-reads/pcawg_analysis/data/processed_data/telomere.rda")