given the possibility to remove temporary files and specify output file

metaclade2

@@ -16,7 +16,6 @@
 trap ctrl_c INT TERM ERR; function ctrl_c() { exit 5; }
 trap 'kill 0' EXIT
 
-
 CMD_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}" )" && pwd)
 CMD_NAME=$(basename "${BASH_SOURCE[0]}")
 SCRIPTS_DIR="${CMD_DIR}"/scripts
@@ -34,6 +33,8 @@ DAMA_OLPMAXDOM=50
 MCLADE_USEDAMA=false
 MCLADE_USESGE=false
 MCLADE_WORKDIR=${PWD}
+MCLADE_OUTFILE=""
+MCLADE_RMTMP=false
 NTHREADS=1
 NJOBS=16
 
@@ -41,11 +42,13 @@ function print_usage() {
     echo -en "\n  USAGE: ${CMD_NAME} -i <input_fasta> -N <name> [options]\n"
     echo -en "\n"
     echo -en "  MANDATORY OPTIONS:\n
+    -N, --name <str>\tDataset/job name and name of the directory used to
+                    \tstore intermediate results\n
     -i, --input <path>\tInput file of AA sequences in FASTA format\n
-    -N, --name <str>\tDataset/job name\n
     " | column -t -s $'\t'
     echo -en "\n"
-    echo -en "  MetaCLADE OPTIONS:\n
+    echo -en "  MetaCLADE2 OPTIONS:\n
+    -o, --output <path>\tOutput file of domain architecture\n
     -a, --arch\tUse DAMA to properly compute domain architectures\n
               \t(useful only for long protein sequences)\n
     -d, --domain-list <str>\tComma-spearated list of Pfam accession numbers of\n
@@ -53,7 +56,8 @@ function print_usage() {
                         \t(e.g., \"PF00875,PF03441\")\n
     -D, --domain-file <path>\tFile that contains the Pfam accession numbers\n
                             \tof the domains to be considered (one per line)\n
-    -W, --work-dir <path>\tWorking directory, where jobs and results are saved\n
+    -W, --work-dir <path>\tWorking directory (default:current directory)\n
+    --remove-temp\tRemove temporary intermediate files, keeping only results and logs\n
     " | column -t -s $'\t'
     echo -en "\n"
     echo -en "  DAMA OPTIONS:\n
@@ -85,8 +89,8 @@ function print_usage() {
 
 
 # retrieve provided arguments
-opts="i:N:ad:D:e:E:W:t:j:hV"
-longopts="input:,name:,arch,domain-list:,domain-file:,evalue-cutoff:,evalue-cutconf:,overlappingAA:,overlappingMaxDomain:,work-dir,threads:,jobs:,help,version,sge,pe:,queue:,time-limit:"
+opts="N:i:o:ad:D:e:E:W:t:j:hV"
+longopts="name:,input:,output:,arch,domain-list:,domain-file:,evalue-cutoff:,evalue-cutconf:,overlappingAA:,overlappingMaxDomain:,work-dir,remove-temp,threads:,jobs:,help,version,sge,pe:,queue:,time-limit:"
 ARGS=$(getopt -o "${opts}" -l "${longopts}" -n "${CMD_NAME}" -- "${@}")
 if [ $? -ne 0 ] || [ $# -eq 0 ]; then # do not change the order of this test!
     print_usage
@@ -96,13 +100,17 @@ eval set -- "${ARGS}"
 
 while [ -n "${1}" ]; do
     case ${1} in
+        -N|--name)
+            shift
+            MCLADE_JOBNAME=${1}
+            ;;
         -i|--input)
             shift
             INPUT_FASTA=${1}
             ;;
-        -N|--name)
+        -o|--output)
             shift
-            MCLADE_JOBNAME=${1}
+            MCLADE_OUTFILE=${1}
             ;;
         -a|--arch)
             MCLADE_USEDAMA=true
@@ -135,6 +143,9 @@ while [ -n "${1}" ]; do
             shift
             MCLADE_WORKDIR=${1}
             ;;
+        --remove-temp)
+            MCLADE_RMTMP=true
+            ;;
         -t|--threads)
             shift
             NTHREADS=${1}
@@ -249,12 +260,15 @@ if [ $? -ne 0 ]; then
 fi
 print_status "MetaCLADE working directory: ${MCLADE_WORKDIR}"
 
+if [ -z "${MCLADE_OUTFILE}" ]; then MCLADE_OUTFILE="${MCLADE_WORKDIR}"/"${MCLADE_JOBNAME}".arch.txt; fi
+print_status "output will be saved to: ${MCLADE_OUTFILE}"
+
 # Create MetaCLADE scripts
 print_status "creating MetaCLADE script/job files"
 python3 "${SCRIPTS_DIR}/mclade_create_jobs.py" \
     -i "${INPUT_FASTA}" -N "${MCLADE_JOBNAME}" ${MCLADE_DOMARG} -W "${MCLADE_WORKDIR}" \
     -e "${MCLADE_EVALUECUTOFF}" -E "${MCLADE_EVALUECUTCONF}" --overlappingAA ${DAMA_OLPAA} --overlappingMaxDomain ${DAMA_OLPMAXDOM} ${MCLADE_DAMAARG} \
-    -j "${NJOBS}"
+    -j "${NJOBS}" >"${MCLADE_WORKDIR}/log/mclade_create_jobs.out" 2>"${MCLADE_WORKDIR}/log/mclade_create_jobs.err"
 
 # Run MetaCLADE scripts (possibly using SGE)
 
@@ -328,6 +342,8 @@ else
     ${PEXEC_CMD} ${f} 2>"${MCLADE_WORKDIR}/log/arch.out" >"${MCLADE_WORKDIR}/log/arch.err"
     cmdret=$?; if (( cmdret != 0 )); then print_error "architecture job failed (exit status: ${cmdret})"; exit 1; fi
 fi
+cp "${MCLADE_WORKDIR}/results/3_arch/${MCLADE_JOBNAME}.arch.txt" "${MCLADE_OUTFILE}"
 print_status "architecture job finished successfully"
 
+if [ "${MCLADE_RMTMP}" = true ]; then print_status "removing temporary files"; rm -r "${MCLADE_WORKDIR}"/{results,jobs,temp}; fi