impute_1000genomes_himem.pbs

impute_1000genomes_himem.pbs

#!/bin/bash

#PBS -M youremailaddress@domain.com
#PBS -m abe
#PBS -S /bin/bash 
#PBS -l nodes=1:ppn=1
#PBS -W x=NODESET:FIRSTOF:FEATURE:opteron:x86
#PBS -l walltime=96:00:00
#PBS -l cput=96:00:00
#PBS -l pmem=8000mb
#PBS -l mem=8000mb
#PBS -j oe

# Otto Valladares, Jacki Buros 
# Version 1.0.pbs
# Created 2010-09-24

# Purpose : 
#	Prepare environment variables for 1000-genomes imputation
# 	Run 1000-genomes imputation per ADGC-organized pipeline

#set current working directory
cd $PBS_O_WORKDIR 

# prefix used for file naming within this study (no spaces)
export prefix=$PREFIX

# settings for this run
export chr=$CHR
if [ -z $GRP ]
then
  # if no group is specified run step1 impute
	step=1
	script="impute_step1.sh"
else 
  # group exists so will run step2 impute script
	export grp=`printf "%02d" $GRP`
	step=2
	script="impute_step2.sh"
fi

# operating directories (note : files copied to OUTDIR at completion)
SRCDIR="$(pwd)"
OUTDIR="$(pwd)"
ERRDIR="$OUTDIR/error_${prefix}_chr${chr}_step${step}_grp${grp}_${JOB_ID}"

# executables 
export MACHBIN="/home/bin/mach1" #CHANGE
export PLINKBIN="/usr/local/plink/plink" #CHANGE
export GAWKBIN="/usr/bin/gawk" #CHANGE
export TARBIN="/bin/tar" #CHANGE

# 1000 genomes files that you downloaded and used tar -zxvf to extract
export REFHAPS="/scratch/username/impute/1000genomes/hap"
export REFSNPS="/scratch/username/impute/snps"

# ** NOTE : scratch dir is working directory where file manipulations / etc can take place.
# This directory will be created & deleted during the operation of the script
# For some clusters, this is ideally located on a node to reduce I/O, but it can be anywhere.
SCRATCHDIR="~/scratch/${prefix}_chr${chr}_step${step}_grp${grp}_${JOB_ID}"

# tar files created / sourced 
# s0file is supposed to exist before running
export s0file="${prefix}_chr${chr}_step0.tar.gz"
export s1file="${prefix}_chr${chr}_step1.tar.gz"
export s2file="${prefix}_chr${chr}_grp${grp}_step2.tar.gz"

# naming conventions used for files in step1 and step2 tarfiles
export s1out="${prefix}_chr${chr}_step1"
export s2out="${prefix}_1000genomes_chr${chr}_grp${grp}"

# error codes indicating success/failure and whether to copy files to $OUTDIR or $ERRDIR
export OK=0
export ERR_DELETE_FILES=1
export ERR_COPY_FILES=2

echo "=========================================================="
echo "Starting on       : $(date)"
echo "Running on node   : $(hostname)"
echo "Current directory : $(pwd)"
echo "Current job ID    : $JOB_ID"
echo "Current job name  : $JOB_NAME"
echo " __ environment variables __"
echo "Scratch dir 	: $SCRATCHDIR"
echo "Study prefix 	: $prefix"
echo "scriptfile 	: $script"
echo "=========================================================="


# create scratch directory
mkdir -p ${SCRATCHDIR}
if [ -d ${SCRATCHDIR} ]; then
	cd ${SCRATCHDIR}
else
	echo "scratchdir ${SCRATCHDIR} does not exist. Exiting"
	exit
fi

# copy source files to scratch directory
cp -a --dereference ${SRCDIR}/${script} .
cp -a --dereference ${SRCDIR}/${s0file} .
if [[ $step -eq 2 ]] ; then
	cp -a --dereference ${SRCDIR}/${s1file} .
fi

# run imputation script -- note : requires script & tar files in working directory
source ${script}
rc=$?

# clean up files 
if [[ $rc -eq $OK ]] ; then
	# create & copy tar file
	if [[ $step -eq 1 ]] ; then
		$TARBIN cfz ${s1file} ${s1out}.*
		mv ${s1file} ${OUTDIR}
		rc=$?
	elif [[ $step -eq 2 ]] ; then
		$TARBIN cfz ${s2file} ${s2out}.*
		mv ${s2file} ${OUTDIR}
		rc=$?
	fi
	# Clean up scratch dir
	if [[ $rc -eq 0 ]] ; then
		echo "=========================================================="
		echo "Finished on       : $(date)"
		echo "=========================================================="
		cd ${OUTDIR}
		rm -rf ${SCRATCHDIR}
		exit
	else
		echo "Unable to copy resulting tar file to $OUTDIR."
		echo "Please ssh to $(hostname) and"
		echo "look at the content of ${SCRATCHDIR}"
		exit
	fi
elif [[ $rc -eq $ERR_DELETE_FILES ]] ; then 
	echo "=========================================================="
	echo "Script exited with errors; scratch dir deleted."
	echo "Finished on       : $(date)"
	echo "=========================================================="
	cd ${OUTDIR}
	rm -rf ${SCRATCHDIR}
	exit
elif [[ $rc -eq $ERR_COPY_FILES ]] ; then
	echo "=========================================================="
	echo "Script exited with errors; copying source files to ${ERRDIR}."
	echo "Finished on       : $(date)"
	echo "=========================================================="
	mkdir -p ${ERRDIR}
	if [ -d ${ERRDIR} ]; then
		mv * ${ERRDIR}
	else
		echo "Unable to copy resulting files to $ERRDIR."
		echo "Please ssh to $(hostname) and"
		echo "look at the content of ${SCRATCHDIR}"
		exit
	fi
	cd ${OUTDIR}
	rm -rf ${SCRATCHDIR}
	exit
fi