gatk_calibrate.scr

#!/bin/sh
#$ -cwd
# -l mem=8G,time=8::

# This step in the pipeline performs calibration of the original input bam file.

GLOBAL="global_config.sh"

if [[ -e $GLOBAL ]]
then
        . $GLOBAL
else
        echo "Global config file not found. Exiting."
        exit 1
fi

USAGE="Usage: $0 -I <Input bam file> -R <Reference fasta> -D <DBSNP file> [-L \"#:#-#\"]"

while getopts I:L:R:D:h o
do      case "$o" in
        I)      INP="$OPTARG";;
        L)      CHR="$OPTARG";;
        R)      REF="$OPTARG";;
        D)      DBSNP="$OPTARG";;
        h)      echo $USAGE
                exit 1;;
        esac
done

if [[ $INP == "" || $REF == "" || $DBSNP == "" ]]
then
        echo $USAGE
        exit 1
fi

# Chromosome number may be specified in eithe of the formats for example:
# chrX or X
# Find out using the samtools idxstats command.

CHR_NAME=`$SAMTOOLS idxstats $INP | grep chr`

if [[ $CHR != "" ]]
then
        if [[ `grep chr $CHR` == "" && -n $CHR_NAME ]]
	then
        	CHR="chr${CHR}"
	fi
fi

$GATK \
 -L $CHR \
 -R $REF \
 --DBSNP $DBSNP \
 -I $INP \
 -T CountCovariates \
 -cov ReadGroupCovariate \
 -cov QualityScoreCovariate \
 -cov CycleCovariate \
 -cov DinucCovariate \
 -recalFile $INP.recal_data.csv

# [-B:mask,VCF sitesToMask.vcf] \