added the possibility to use SGE for search/filter, needs testing

metaclade2

@@ -34,12 +34,6 @@ MCLADE_WORKDIR=${PWD}
 NTHREADS=1
 NJOBS=16
 
-PEXEC_CMD="parallel --halt now,fail=1 -j ${NJOBS}"
-if ! command -v parallel >/dev/null 2>&1; then
-    print_warning "cannot find GNU parallel, all jobs will be run sequentially"
-    PEXEC_CMD="/usr/bin/env bash --"
-fi
-
 function print_usage() {
     echo -en "\n  USAGE: ${CMD_NAME} -i <input_fasta> [options]\n"
     echo -en "\n"
@@ -143,6 +137,13 @@ done
 
 # Input parameters validation
 
+PEXEC_CMD="parallel --halt now,fail=1 -j ${NTHREADS} ::::"
+if ! command -v parallel >/dev/null 2>&1; then
+    print_warning "cannot find GNU parallel, all jobs will be run sequentially"
+    NTHREADS=1
+    PEXEC_CMD="/usr/bin/env bash --"
+fi
+
 if [ -z "${INPUT_FASTA}" ] || [ ! -f ${INPUT_FASTA} ]; then
     print_error "-i|--input file is missing or does not exist: ${INPUT_FASTA}"
     exit 1
@@ -181,24 +182,83 @@ fi
 
 check_cmds "python3"
 
+# Validate SGE parameters
+if [ "${MCLADE_USESGE}" = true ] ; then
+    if [ -z ${SGE_PENAME} ]; then
+        print_error "you must set a parallel environment name with -P|--parallel-env option along with the --SGE option"
+        exit 1
+    fi
+    if [ ! -z ${SGE_QUEUE} ]; then
+        SGE_QUEUEARG="-q ${SGE_QUEUE}"
+    fi
+    if [ ! -z ${SGE_TIMELIM} ]; then
+        SGE_TIMELIMARG="-l h_rt=${SGE_TIMELIM}"
+    fi
+fi
+
 # Create Working directory
 MCLADE_WORKDIR=${MCLADE_WORKDIR}"/"${MCLADE_JOBNAME}
-mkdir -p ${MCLADE_WORKDIR}
+mkdir -p "${MCLADE_WORKDIR}" "${MCLADE_WORKDIR}/log"
 if [ $? -ne 0 ]; then
     print_error "cannot create/access MetaCLADE working directory: ${MCLADE_WORKDIR}"
     exit 1
 fi
 print_status "MetaCLADE working directory: ${MCLADE_WORKDIR}"
 
+# Create MetaCLADE scripts
 print_status "creating MetaCLADE script/job files"
 python3 "${SCRIPTS_DIR}/mclade_create_jobs.py" -i "${INPUT_FASTA}" -N "${MCLADE_JOBNAME}" ${MCLADE_DOMARG} -W "${MCLADE_WORKDIR}" -e "${MCLADE_EVALUECUTOFF}" -E "${MCLADE_EVALUECUTCONF}" -j "${NJOBS}"
 
+# Possibly run MetaCLADE scripts in a SGE environment
 if [ "${MCLADE_USESGE}" = true ] ; then
-    # run search jobs
+    # submit search jobs
+    print_status "submitting search jobs"
+    pidarr=()
     for i in $(seq 1 ${NJOBS}); do
         f="${MCLADE_WORKDIR}/jobs/1_search/${MCLADE_JOBNAME}_${i}.sh"
+        # run a qsub job for each non-empty script
+        if [ -f "${f}" ] && [ -s "${f}" ]; then
+            qsub $SGE_QUEUEARG $SGE_TIMELIMARG -e "${MCLADE_WORKDIR}/log/search_${i}.err" -o "${MCLADE_WORKDIR}/log/search_${i}.out" -cwd -sync yes -N ${MCLADE_JOBNAME} -pe ${SGE_PENAME} ${NTHREADS} -b y ${PEXEC_CMD} ${f} &
+            pid=$!
+            pidarr[$i]=$pid
+            ((i++))
+        fi
+    done
+    # wait search jobs to finish
+    print_status "waiting the search jobs to finish"
+    for i in ${!pidtab[@]}; do
+        wait ${pidtab[${i}]}
+        ret=$?
+        if ((ret != 0)); then
+            echo "MetaCLADE search job ${i} failed (exit status: ${ret})"
+            exit 1
+        fi
+    done
+    echo "search jobs finished successfully"
+
+    # submit filter jobs
+    print_status "submitting filter ijobs"
+    unset pidarr
+    pidarr=()
+    for i in $(seq 1 ${NJOBS}); do
+        f="${MCLADE_WORKDIR}/jobs/1_filter/${MCLADE_JOBNAME}_${i}.sh"
+        # run a qsub job for each non-empty script
         if [ -f "${f}" ] && [ -s "${f}" ]; then
-            echo "qsub -sync yes -b y ${f}"
+            qsub $SGE_QUEUEARG $SGE_TIMELIMARG -e "${MCLADE_WORKDIR}/log/filter_${i}.err" -o "${MCLADE_WORKDIR}/log/filter_${i}.out" -cwd -sync yes -N ${MCLADE_JOBNAME} -pe ${SGE_PENAME} ${NTHREADS} -b y ${PEXEC_CMD} ${f} &
+            pid=$!
+            pidarr[$i]=$pid
+            ((i++))
+        fi
+    done
+    # wait search jobs to finish
+    print_status "waiting the filter jobs to finish"
+    for i in ${!pidtab[@]}; do
+        wait ${pidtab[${i}]}
+        ret=$?
+        if ((ret != 0)); then
+            echo "MetaCLADE filter job ${i} failed (exit status: ${ret})"
+            exit 1
         fi
     done
+    echo "search jobs finished successfully"
 fi