"Strategy" for generating shuffled BED files while preventing overlapping records in the shuffled output.

shuffle-non-overlapping.sh

tries=0

while true; 
do
  tries=$((tries+1))
  echo "attempt number $tries"

  # try a shuffle
  bedtools shuffle -i foo.bed -g human.hg19.genome > foo.shuffled.bed
  
  # a. sort the shuffle by chrom and start position so that we can use 
  #    bedtools merge to test for overlaps
  # b. the -n option in bedtools merge reports how many intervals in the 
  #    original file are represented in the merged blocks.  As such, if
  #    all of these counts are == 1, then there were no overlapping intervals
  sort -k1,1 -k2,2n foo.shuffled.bed | bedtools merge -i - -n > foo.shuffled.merged.bed
  
  # test to see if there were any overlapping intervals
  has_overlaps=`awk '$4 > 1' foo.shuffled.merged.bed | wc -l`

  # if there were not, we can quit and use foo.shuffled.merged.bed
  if [ $has_overlaps == "0" ]
    then break;
  fi
done