...
 
......@@ -121,7 +121,7 @@ declare _arg_ramdisk=40G
_arg_blastdir='.'
_arg_executable='blastx'
_arg_test=off
_arg_compress=off
_arg_compress=on
print_help ()
{
......@@ -577,11 +577,6 @@ if [[ ! $allowed_executables =~ (^|[[:space:]])"$_arg_executable"($|[[:space:]])
#else
error "$_arg_executable ought to be one of [$allowed_executables]"
fi
BLASTEXE=$(which $_arg_executable)
if [ -z $BLASTEXE ] && [[ $_arg_test == "off" ]]; then
error "'$BLASTEXE' : not found, please load an appropriate module before restarting."
exit
fi
### which is the reference directory size?
_arg_ramdisk=$(du -shL --block-size=1M "$_arg_database" | cut -f1 )M
......@@ -764,7 +759,6 @@ $parallel "$srun" "$cmdfile" ::: $(find $(pwd) -type f -name 'group*.fasta')
wait
set -x
n_unfinished_files=$(comm -3 <(cd output && find .| grep -o '[0-9]*' |sort ) <(cd scratch && find . | grep -o '[0-9]*' |sort )|wc -l)
if [ $n_unfinished_files -eq 0 ] && [[ $_arg_compress == "on" ]] && [ $XMLOUT -eq 1 ]; then
# shrink the alloction, such that only the minimum necessary is accounted for
......@@ -801,6 +795,7 @@ elif [ $n_unfinished_files -eq 0 ] && [[ $_arg_compress == "on" ]] && [ $OUTOUT
# write anything to the output file
for split_file in ./output/group_*gz; do
zcat $split_file >> $outfile
rm $split_file
done
pigz -p 16 $outfile &
ENDC=$(date +%s.%N)
......
function stage_in_writer() {
cat <<EOF > $stagefile
#!/bin/bash
target=/localscratch/$SLURM_JOB_ID/ramdisk
cd \$target
for fname in \$(find ${DATABASEPATH} -type f ); do
suffix=\${fname#*.}
outfile=db.\${suffix}
dd bs=4048 if=\$fname of=\$outfile &
done
cd -
EOF
}