first modularization: splitter script now outsourced and addressed by variable

blast/parallel_BLAST/parallel_BLAST

@@ -532,49 +532,6 @@ if [[ ! $SCRIPT == /* ]]; then
    SCRIPT="$PWD/$SCRIPT";
 fi
 
-### check if this script is on node by checking env-variable $LSB_JOBID, else send it to LSF with given parameters and exit
-SPLITUPSCRIPT=$(cat <<'EOF'
-# dummy line to introduce a line break
-import pip
-# will take little time, if dependency is already satisfied
-pip.main(['install', 'biopython'])
-
-from Bio import SeqIO
-import sys
-import os
-
-fasta_file = sys.argv[1]
-# number of sequences per scratch output file
-nlines     = int(sys.argv[2])
-
-print('Going to split: >%s<' % fasta_file)
-print('Going to split in %s lines per output file.' % nlines)
-
-record_iter = SeqIO.parse(open(fasta_file),'fasta')
-group = 0
-batch = list()
-for pos, entry in enumerate(record_iter):
-    if pos == 0:
-        group += 1
-        filename = 'group_%s.fasta' % group
-        handle = open(os.path.join('scratch', filename), 'w')
-    if (pos % nlines == 0 and pos != 0):
-        count = SeqIO.write(batch, handle, 'fasta')
-        #print('Wrote %s records to %s' % (count, filename))
-        handle.close()
-        batch = list()
-        group += 1
-        filename = 'group_%s.fasta' % group
-        handle = open(os.path.join('scratch', filename), 'w')
-    batch.append(entry)
-# take care of the rest
-count = SeqIO.write(batch, handle, 'fasta')
-print('Wrote %s records to %s' % (count, filename))
-handle.close()
-    
-EOF
-)
-
 ### check if this script is on node by checking env-variable $SLURM_JOB_ID, else send it to SLURM with given parameters and exit
 if [ -z "$SLURM_JOB_ID" ]; then
     submit_call="sbatch --no-requeue -o ${JOBTAG}_%j.out -J $JOBTAG -p $_arg_queue -A $_arg_assoc -t $_arg_runlimit -N $_arg_nodes -n $((64 * $_arg_nodes / $_arg_blast_threads)) --mem=$_arg_mem --ramdisk=${_arg_ramdisk} -c $_arg_blast_threads --reservation=$_arg_reservation $SCRIPT --partition $_arg_queue --account $_arg_assoc --nodes $_arg_nodes --time $_arg_runlimit --reservation=$_arg_reservation  --threads $_arg_blast_threads --splitup $_arg_splitup_per_queryfile --blastparams=\"$BLASTPARAMS\" --executable=$_arg_executable $FASTA $DATABASE"
@@ -636,7 +593,7 @@ if [ ! -d "$WORKDIR/$SPLITFILEDIR" ]; then
     mkdir -p "$WORKDIR/output" || exit 1;
     cd "$WORKDIR"
     echo "executing scratch generator on $FASTA ($_arg_splitup_per_queryfile entries per file)"
-    eval "python -c \"$SPLITUPSCRIPT\" $FASTA $_arg_splitup_per_queryfile" &     # splitup queryfile
+    eval "${SCRIPT_PATH}/splitter.py $FASTA $_arg_splitup_per_queryfile" &     # splitup queryfile
     PID=$!
     queue $PID
 fi