#!/bin/bash
#
# bash unix shell script for dividing a sample into subsamples and 
# performing imputation with Beagle on each subsample separately.
#
# paste.jar and cut.jar programs are available from the BEAGLE Utilities web page
# (http://www.stat.auckland.ac.nz/~bbrowning/beagle_utilities/utilities.html)

sample="?"	# replace ? with name of sample Beagle file
ref="?"		# replace ? with name of reference panel Beagle file
missing="?"	# replace ? with missing allele code
markers="?"	# replace ? with name of markers file with markers in reference panel

sampleCol="?"	# replace ? with number of columns (including first two columns) in sample Beagle file
windowSize="?"	# replace ? of number of columns of allele data to use per subsample (e.g. 500)

start="3"
let "end=start+windowSize-1"
z="10"

while [ "${end}" -lt "${sampleCol}" ]
do
  echo "start=${start}  end=${end}  subsample.index=${z}"
  cat ${sample} | java -jar cut.jar 1:2 ${start}:${end} > ${sample}.${z}
  java -jar beagle.jar phased=${ref} unphased=${sample}.${z} missing=${missing} markers=${markers} out=out.${z}
  let "start=end+1"
  let "end=start+windowSize-1"
  let "z=z+1"
done

echo "start=${start}  end=${sampleCol}  subsample.index=${z}"
cat ${sample} | java -jar cut.jar 1:2 ${start}:${sampleCol} > ${sample}.${z}
java -jar beagle.jar phased=${ref} unphased=${sample}.${z} missing=${missing} markers=${markers} out=out.${z}

java -jar paste.jar 2 out.[1-9][0-9].${sample}.[1-9][0-9].phased.gz | gzip > ${sample}.phased.gz
java -jar paste.jar 3 out.[1-9][0-9].${sample}.[1-9][0-9].gprobs.gz | gzip > ${sample}.gprobs.gz
java -jar paste.jar 3 out.[1-9][0-9].${sample}.[1-9][0-9].dose.gz | gzip > ${sample}.dose.gz
java -jar paste.jar 1 out.[1-9][0-9].${sample}.[1-9][0-9].r2 > ${sample}.r2
cat out.[1-9][0-9].log > ${sample}.log

rm ${sample}.[1-9][0-9]
rm out.[1-9][0-9].*