#!/bin/bash

# Parses UCSC Chromosome Band table into a Circos karyotype file. 
#
# This script will only work on the Chromosome Band table - it will not 
# work on an 'Assembly' or 'Scaffold' table. Many organisms do not have
# a Chromosome Band table, and for these you will need to 
#
# parse.karyotype chromosome.band.hg19.txt hs > karyotype.hg19.txt
#
# The prefix, e.g. hs, will be used to label the chromosomes (e.g. hs1, hs2 ... )
# By default, "chr" is used.
#
# To download data tables, see http://genome.ucsc.edu/cgi-bin/hgTables

file=$1
prefix=$2

if [ ! -n "$file" ]
then
echo "Plese specify the UCSC karyotype table file"
exit
fi

if [ ! -n "$prefix" ]
then
prefix="chr"
fi

# chromosomes
cat $file | grep -v ^\# | tr A-Z a-z | sort +0 -1 +2rn -3 | sort -u -k 1,1 | sed 's/\t/ /g' | tr -s " " | sed 's/chr//' | awk '{print $1,$0}' | awk -v prefix=$2 '{print "chr -",prefix$2,$1,0,$4,"chr"$1}' | sort -n +3 -4 | sed 's/x/X/g' | sed 's/y/Y/g'

# bands
cat $file  | grep -v ^\# | tr A-Z a-z | sed 's/\t/ /g' | tr -s " " | sed "s/chr/$prefix/" | awk '{print $1,$0}' | sed 's/chr//' | awk '{print "band",$2,$5,$5,$3,$4,$6}' | sort +1 -2 +4n -5 | sed 's/x/X/g' | sed 's/y/Y/g'