phaplo

#!/bin/bash

#Run haplosaurus in parallel


if [ "$1" == "-h" ]; then
  echo "Usage: `basename $0` [my.vcf]"
  exit 0
fi


#make sure there is an inputfile
if [ $# -eq 0 ]
  then
    echo "No arguments supplied"
    echo "Use -h for info"
    exit 1
fi


#grab the header
head -n 10000 $1 | grep "^#" >header
#grab the non header lines
grep -v "^#" $1 >variants
#split into chunks with 1000 lines
split --number=10 variants

wait

#put the headers back on
cat header xaa > xaa.header &
cat header xab > xab.header &
cat header xac > xac.header &
cat header xad > xad.header &
cat header xae > xae.header &
cat header xaf > xaf.header &
cat header xag > xag.header &
cat header xah > xah.header &
cat header xai > xai.header &
cat header xaj > xaj.header &

wait

#run haplo
haplo -i xaa.header -o xaa.haplo.tsv --cache --port 3337 --dir /mnt/home/stephen/.vep/hg19/ &
haplo -i xab.header -o xab.haplo.tsv --cache --port 3337 --dir /mnt/home/stephen/.vep/hg19/ &
haplo -i xac.header -o xac.haplo.tsv --cache --port 3337 --dir /mnt/home/stephen/.vep/hg19/ &
haplo -i xad.header -o xad.haplo.tsv --cache --port 3337 --dir /mnt/home/stephen/.vep/hg19/ &
haplo -i xae.header -o xae.haplo.tsv --cache --port 3337 --dir /mnt/home/stephen/.vep/hg19/ &
haplo -i xaf.header -o xaf.haplo.tsv --cache --port 3337 --dir /mnt/home/stephen/.vep/hg19/ &
haplo -i xag.header -o xag.haplo.tsv --cache --port 3337 --dir /mnt/home/stephen/.vep/hg19/ &
haplo -i xah.header -o xah.haplo.tsv --cache --port 3337 --dir /mnt/home/stephen/.vep/hg19/ &
haplo -i xai.header -o xai.haplo.tsv --cache --port 3337 --dir /mnt/home/stephen/.vep/hg19/ &
haplo -i xaj.header -o xaj.haplo.tsv --cache --port 3337 --dir /mnt/home/stephen/.vep/hg19/ &

wait

cat xa*.haplo.tsv  > $1.haplo.tsv
cat /mnt/home/stephen/Apps/bin/haplo.header $1.haplo.tsv


rm -f header variants xa*.haplo.vcf *.header