Get splitters and discordants

split_and_discord_it.sh

BAM=$1

OUT_DIR=$2

LUMPY_HOME=$HOME/src/lumpy-sv
SAMBAMBA=$HOME/bin/sambamba
SAMBLASTER=$HOME/bin/samblaster
BAMLIBS=$LUMPY_HOME/scripts/bamkit/bamlibs.py
SAMTOBAM="$SAMBAMBA view -S -f bam -l 0"
SAMSORT="$SAMBAMBA sort -m 1G --tmpdir "
PAIREND_DISTRO=$LUMPY_HOME/scripts/pairend_distro.py
BAMGROUPREADS=$LUMPY_HOME/scripts/bamkit/bamgroupreads.py
BAMFILTERRG=$LUMPY_HOME/scripts/bamkit/bamfilterrg.py
MAX_SPLIT_COUNT=2
MIN_NON_OVERLAP=20

if [ ! -d "$OUT_DIR" ]; then
    mkdir $OUT_DIR
fi

BAM_BASE=`basename $BAM .bam`
OUTPUT=`basename $BAM`.vcf
OUTBASE=`basename "$OUTPUT"`

# make pipes
TEMP_DIR=$OUT_DIR/$BAM_BASE
mkdir -p $TEMP_DIR/spl $TEMP_DIR/disc
mkfifo $TEMP_DIR/spl_pipe
mkfifo $TEMP_DIR/disc_pipe

echo "$PYTHON $BAMGROUPREADS \
    -i $BAM \
| $SAMBLASTER \
    --acceptDupMarks \
    --excludeDups \
    --addMateTags \
    --maxSplitCount $MAX_SPLIT_COUNT \
    --minNonOverlap $MIN_NON_OVERLAP \
    --splitterFile $TEMP_DIR/spl_pipe \
    --discordantFile $TEMP_DIR/disc_pipe "

$PYTHON $BAMGROUPREADS \
    -i $BAM \
| $SAMBLASTER \
    --acceptDupMarks \
    --excludeDups \
    --addMateTags \
    --maxSplitCount $MAX_SPLIT_COUNT \
    --minNonOverlap $MIN_NON_OVERLAP \
    --splitterFile $TEMP_DIR/spl_pipe \
    --discordantFile $TEMP_DIR/disc_pipe  \
> /dev/null &

$SAMTOBAM $TEMP_DIR/spl_pipe \
| $SAMSORT $TEMP_DIR/spl \
    -o $TEMP_DIR/$BAM_BASE.splitters.bam  \
    /dev/stdin &

$SAMTOBAM $TEMP_DIR/disc_pipe \
| $SAMSORT $TEMP_DIR/disc \
    -o $TEMP_DIR/$BAM_BASE.discordants.bam \
    /dev/stdin

wait
rm $TEMP_DIR/spl_pipe
rm $TEMP_DIR/disc_pipe
rmdir $TEMP_DIR/spl
rmdir $TEMP_DIR/disc