attempt to fix macOS problem with getopt

profileview-build

@@ -34,14 +34,8 @@ HHBLITS_ARG_EVALUE=1e-10
 PVLIB_FORCE=false
 PVLIB_PATH=""
 PVLIB_DOMID=""
-NTHREADS=2
-NJOBS=8
-
-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
+NTHREADS=4
+NJOBS=1
 
 function print_usage() {
     echo -en "\n  USAGE: ${CMD_NAME} -i <input_fasta> -d <hhblits_db> -n <lib_name> [options]\n"
@@ -55,20 +49,27 @@ function print_usage() {
     -n, --lib-name <name>\tName of ProfileView library\n" | column -t -s $'\t'
     echo -en "\n"
     echo -en "  OTHER OPTIONS:\n
-    --force\tForce the construction of ProfileView's library even if the directory already exists\n
+    -f, --force\tForce the construction of ProfileView's library even if the directory already exists\n
            \tpossibly overwriting previously generated files\n
-    -t, --threads <num>\tNumber of threads for each hhblits job (default:2)\n
-    -j, --max-jobs <num>\tNumber of parallel jobs (default:8)\n
+    -t, --threads <num>\tNumber of threads for each hhblits job (default:${NTHREADS})\n
+    -j, --max-jobs <num>\tNumber of parallel jobs (default:${NJOBS})\n
     -h, --help\tPrint this help message\n
     -V, --version\tPrint version\n" | column -t -s $'\t'
     echo -en "\n"
+    echo -en "  Long options (e.g., --input) are supported only on Linux systems\n\n"
 }
 
 # retrieve provided arguments
-opts="i:D:d:n:t:j:hV"
+opts="i:D:d:n:ft:j:hV"
 longopts="input:,domain-id:,db:,lib-name:,force,threads:,max-jobs:,help,version,cov:,qid:,max-id:,evalue:,iter:"
-ARGS=$(getopt -o "${opts}" -l "${longopts}" -n "${CMD_NAME}" -- "${@}")
-if [ $? -ne 0 ] || [ $# -eq 0 ]; then # do not change the order of this test!
+if [ check_system = "Linux" ]; then
+    ARGS=$(getopt -o "${opts}" -l "${longopts}" -n "${CMD_NAME}" -- "${@}")
+    retcode=$?
+else
+    ARGS=$(getopt "${opts}" $*)
+    retcode=$?
+fi
+if [ ${retcode} -ne 0 ] || [ $# -eq 0 ]; then # do not change the order of this test!
     print_usage
     exit 1
 fi
@@ -93,7 +94,7 @@ while [ -n "${1}" ]; do
             shift
             PVLIB_PATH=${1}
             ;;
-        --force)
+        -f|--force)
             PVLIB_FORCE=true
             ;;
         -t|--threads)
@@ -184,6 +185,12 @@ fi
 
 check_cmds "hhblits" "reformat.pl" "hmmbuild" "python3"
 
+PEXEC_CMD="parallel --halt now,fail=1 -j ${NJOBS}"
+if ! command -v parallel >/dev/null 2>&1; then
+    print_warning "cannot find GNU parallel, all jobs will be run sequentially"
+    PEXEC_CMD="/usr/bin/env bash --"
+fi
+
 # Create ProfileView library directory
 PVLIB_NAME=$(basename "${PVLIB_PATH}")
 PVLIB_PREFIX="$(dirname "${PVLIB_PATH}")"/"${PVLIB_NAME}"

profileview-common

@@ -17,7 +17,7 @@
 # Common functions and definitions to be used in ProfileView
 PV_NAME="ProfileView"
 PV_VERSION='1.0'
-PV_DATE='20190726'
+PV_DATE='20200513'
 
 function abspath() { echo "$(cd "$(dirname "$1")"; pwd -P)/$(basename "$1")"; }
 
@@ -72,4 +72,16 @@ function check_files() {
     return 0
 }
 
+function check_system() {
+    machine="UNKNOWN"
+    case "$(uname -s)" in
+        Linux*)  machine=Linux;;
+        Darwin*) machine=Mac;;
+        CYGWIN*) machine=Cygwin;;
+        MINGW*)  machine=MinGw;;
+        *)       machine="UNKNOWN:$(uname -s)";;
+    esac
+    echo ${machine}
+}
+
 

profileview-motif

@@ -33,13 +33,7 @@ PV_OUTTREE=""
 PV_SEQDESC=""
 PV_OUTDIR="out_$(date +%Y%m%d_%H%M%S)"
 PV_TMPDIR=""
-NJOBS=8
-
-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
+NJOBS=4
 
 function print_usage() {
     echo -en "\n  USAGE: ${CMD_NAME} -l <profileview_lib> -m <model-list> [options]\n"
@@ -51,22 +45,29 @@ function print_usage() {
     echo -en "\n"
     echo -en "  OTHER OPTIONS:\n
     -o, --out-dir  <name>\tPrefix of output files (default: out_<current-date>)\n
-    --temp-dir     <name>\tTemporary result directory\n
-    -j, --max-jobs  <num>\tNumber of parallel jobs (default:8)\n
+    -W, --temp-dir <name>\tTemporary result directory\n
+    -j, --max-jobs <num>\tNumber of parallel jobs (default:${NJOBS})\n
     -h, --help\tPrint this help message and exit\n
     -V, --version\tPrint version and exit\n
     " | column -t -s $'\t'
     echo -en "\n"
+    echo -en "  Long options (e.g., --input) are supported only on Linux systems\n\n"
 }
 
 # retrieve provided arguments
 
-opts="l:m:o:j:hV"
-longopts="lib:,models:,out-dir:,temp-dir:,max-jobs:,help,version"
-ARGS=$(getopt -o "${opts}" -l "${longopts}" -n "${CMD_NAME}" -- "${@}")
-if [ $? -ne 0 ] || [ $# -eq 0 ]; then # the order of this tests is important!
+opts="l:m:o:W:t:hV"
+longopts="lib:,models:,out-dir:,temp-dir:,threads:,help,version"
+if [ check_system = "Linux" ]; then
+    ARGS=$(getopt -o "${opts}" -l "${longopts}" -n "${CMD_NAME}" -- "${@}")
+    retcode=$?
+else
+    ARGS=$(getopt "${opts}" $*)
+    retcode=$?
+fi
+if [ ${retcode} -ne 0 ] || [ $# -eq 0 ]; then
     print_usage
-    exit 2
+    exit 1
 fi
 eval set -- "${ARGS}"
 
@@ -85,11 +86,11 @@ while [ -n "${1}" ]; do
             shift
             PV_OUTDIR=${1}
             ;;
-        --temp-dir)
+        -W|--temp-dir)
             shift
             PV_TMPDIR=${1}
             ;;
-        -j|--max-jobs)
+        -t|--threads)
             shift
             NJOBS=${1}
             ;;
@@ -114,6 +115,12 @@ done
 check_cmds "hhalign" "python3"
 check_pymodules "Bio" "weblogolib" "BitVector"
 
+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
+
 # Input parameters validation
 
 if [ -z "${PV_LIBPATH}" ]; then

profileview-tree

@@ -34,13 +34,7 @@ PV_OUTPREFIX="out"
 PV_TMPDIR=""
 PV_CVAR=0.99
 PV_KBEST=3
-NJOBS=8
-
-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
+NJOBS=4
 
 function print_usage() {
     echo -en "\n  USAGE: ${CMD_NAME} -i <input_fasta> -l <profileview_lib> [options]\n"
@@ -56,21 +50,28 @@ function print_usage() {
     -s, --seq-desc <name>\tInput sequence descriptor file, that is a CSV file containing the follwing fileds:\n
                          \t<sequence_id>,<function_id>,<family_id>,<sequence_length>\n
     -p, --prefix <name>\tPrefix of output files (default: ${PV_OUTPREFIX})\n
-    --temp-dir <name>\tTemporary result directory\n
-    -j, --max-jobs <num>\tNumber of parallel jobs (default: ${NJOBS})\n
+    -W, --temp-dir <name>\tTemporary result directory\n
+    -t, --threads <num>\tNumber of parallel jobs (default: ${NJOBS})\n
     -h, --help\tPrint this help message\n
     -V, --version\tPrint version\n" | column -t -s $'\t'
     echo -en "\n"
+    echo -en "  Long options (e.g., --input) are supported only on Linux systems\n\n"
 }
 
 # retrieve provided arguments
 
-opts="i:l:k:s:p:j:hV"
-longopts="input:,lib:,k-best:,seq-desc:,prefix:,temp-dir:,max-jobs:,cvar:,help,version"
-ARGS=$(getopt -o "${opts}" -l "${longopts}" -n "${CMD_NAME}" -- "${@}")
-if [ $? -ne 0 ] || [ $# -eq 0 ]; then # the order of this tests is important!
+opts="i:l:k:s:p:W:t:hV"
+longopts="input:,lib:,k-best:,seq-desc:,prefix:,temp-dir:,threads:,cvar:,help,version"
+if [ check_system = "Linux" ]; then
+    ARGS=$(getopt -o "${opts}" -l "${longopts}" -n "${CMD_NAME}" -- "${@}")
+    retcode=$?
+else
+    ARGS=$(getopt "${opts}" $*)
+    retcode=$?
+fi
+if [ ${retcode} -ne 0 ] || [ $# -eq 0 ]; then # do not change the order of this test!
     print_usage
-    exit 2
+    exit 1
 fi
 eval set -- "${ARGS}"
 
@@ -93,14 +94,18 @@ while [ -n "${1}" ]; do
             shift
             PV_OUTPREFIX=${1}
             ;;
-        --temp-dir)
+        -W|--temp-dir)
             shift
             PV_TMPDIR=${1}
             ;;
-        -j|--max-jobs)
+        -t|--threads)
             shift
             NJOBS=${1}
             ;;
+        -k|--k-best)
+            shift
+            PV_KBEST=${1}
+            ;;
         --cvar)
             shift
             PV_CVAR=${1}
@@ -165,10 +170,14 @@ if ! [[ "${PV_CVAR}" =~ ^(0(\.[0-9]+)?|\.[0-9]+|1(\.0+)?)$ ]] ; then
     PV_CVAR=0.99
 fi
 
-
 check_cmds "hmmsearch" "python3" "Rscript"
 check_pymodules "ete3" "numpy"
-#check_files "${SCRIPTS_DIR}"/{createHHdict.py,createHmmerDict.py,hh_utils.py,pv_utils.py}
+
+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
 
 # Create temp working directory
 if [ -z "${PV_TMPDIR}" ]; then
@@ -244,7 +253,7 @@ print_status "filtering sequences"
 PV_SEQDESC="${PV_TMPDIR}/sequences.filtered.csv"
 awk '/^#/{next} !x[$3]++{OFS=",";print $3,$6,$5,$4}' "${PV_SCOREFILE}" >"${PV_SEQDESC}" 2>/dev/null
 
-print_status "building representation space"
+print_status "building representation space (using k=${PV_KBEST})"
 python3 "${SCRIPTS_DIR}"/generateFeatures.py --seq-list "${PV_SEQDESC}" --hmm-list "${PV_LIBDIR}/${PV_LIBNAME}.models.list" --scores "${PV_SCOREFILE}" --prefix "${PV_TMPDIR}"/out -n 20 -k "${PV_KBEST}" 2>>"${PV_LOGFILE}"
 if [ $? -ne 0 ]; then
     print_error "error during feature generation, see log: ${PV_LOGFILE}"