Rollup for java clustering

workflow/qhtcp-workflow

@@ -348,14 +348,11 @@ backup() {
   debug "Running: ${FUNCNAME[0]} $*"
   for f in "$@"; do
     [[ -e $f ]] || continue
-    count=1
+    declare count=1
     while [[ -e $f.bk.$count ]]; do
       ((count++))
     done
-    echo "Backing up $f to $f.bk.$count"
-    debug "rsync -a $f $f.bk.$count"
-    rsync -a "$f" "$f.bk.$count" &&
-    rm -rf "$f"
+    execute mv "$f" "$f.bk.$count"
   done
 }
 
@@ -827,14 +824,14 @@ init_project() {
         u)
           echo "Updating project from template"
           echo "Only files that are newer in the template will be overwritten"
-          if rsync --archive --update "$QHTCP_TEMPLATE_DIR"/ "$QHTCP_RESULTS_DIR"; then
+          if execute rsync --archive --update "$QHTCP_TEMPLATE_DIR"/ "$QHTCP_RESULTS_DIR"; then
             echo "Project updated with template"
           fi
           ;;
         b)
           if backup "$QHTCP_RESULTS_DIR"; then
             mkdir "$QHTCP_RESULTS_DIR"
-            rsync --archive "$QHTCP_TEMPLATE_DIR"/ "$QHTCP_RESULTS_DIR"
+            execute rsync --archive "$QHTCP_TEMPLATE_DIR"/ "$QHTCP_RESULTS_DIR"
             echo "Created new project at $QHTCP_RESULTS_DIR"
           fi
           ;;
@@ -850,10 +847,8 @@ init_project() {
     done
   else
     echo "Creating project results dir at $QHTCP_RESULTS_DIR"
-    debug "mkdir $QHTCP_RESULTS_DIR"
-    mkdir "$QHTCP_RESULTS_DIR"
-    debug "rsync --archive $QHTCP_TEMPLATE_DIR/ $QHTCP_RESULTS_DIR"
-    rsync --archive "$QHTCP_TEMPLATE_DIR"/ "$QHTCP_RESULTS_DIR"
+    execute mkdir "$QHTCP_RESULTS_DIR"
+    execute rsync --archive "$QHTCP_TEMPLATE_DIR"/ "$QHTCP_RESULTS_DIR"
   fi
 
   study_info
@@ -1279,8 +1274,6 @@ module qhtcp
 #
 # TODO 
 #
-# * Suggest renaming StudiesQHTCP to something like qhtcp qhtcp_output or output
-# * Move (hide) the study template somewhere else
 # * StudiesArchive should be smarter:
 #   * Create a database with as much information as possible
 #   * Write a function that easily loads and parses databse into easy-to-use variables
@@ -1366,13 +1359,10 @@ qhtcp() {
 
   # Run R interactions script on all studies
   for study in "${STUDIES[@]}"; do
-    read -r num sd dir <<< "$study"
-    [[ -d $dir/zscores ]] ||
-      execute mkdir "$dir/zscores"
-    [[ -d $dir/zscores/qc ]] ||
-      execute mkdir "$dir/zscores/qc"
-    r_interactions "$num" "$sd"
+    read -r num sd _ <<< "$study"
+    r_interactions "$num" "$sd" & # run in parallel, catch with wait below
   done \
+  && wait -n \
   && remc \
   && gtf \
   && gta
@@ -1750,8 +1740,9 @@ r_interactions() {
   declare script="$APPS_DIR/r/interactions.R"
   declare out_dir="${6:-"$QHTCP_RESULTS_DIR/Exp$1/zscores"}"
 
-  [[ -d $out_dir ]] && backup "$out_dir"
-  mkdir "$out_dir"
+  backup "$out_dir"
+  execute mkdir "$out_dir"
+  execute mkdir "$out_dir/qc"
 
   execute "$RSCRIPT" "$script" \
     "$1" \
@@ -1850,7 +1841,7 @@ wrapper java_extract
 # @exitcode 1 if expected output file does not exist
 java_extract() {
   debug "Running: ${FUNCNAME[0]}"
-  classpath="$APPS_DIR/java/javaExtract.jar"
+  classpath="$APPS_DIR/java/weka-clustering.jar"
 
   output_file="${1:-$QHTCP_RESULTS_DIR}/REMcRdy_lm_only.csv-finalTable.csv"
 
@@ -2337,7 +2328,7 @@ main() {
 
   parse_input "$@" # parse arguments with getopt
 
-  # ((DEBUG)) && declare -p
+  ((DEBUG)) && declare -p
 
   interactive_header "$@"
 
@@ -2347,8 +2338,8 @@ main() {
   # fi
   
   for i in "${!PROJECTS[@]}"; do
-    if ! sanitize_pn "${PROJECTS[i]}"; then
-      echo "Project name ${PROJECTS[i]} is invalid"
+    if ! sanitize_pn "${PROJECTS[$i]}"; then
+      echo "Project name ${PROJECTS[$i]} is invalid"
       echo "Enter a replacement"
       ask_pn && unset "PROJECTS[i]" && PROJECTS+=("${ADD_PROJECTS[@]}")
     fi
@@ -2356,15 +2347,13 @@ main() {
 
   # Exclude modules from --exclude
   for i in "${!MODULES[@]}"; do
-    [[ " ${EXCLUDE_MODULES[*]} " =~ [[:space:]]${MODULES[i]}[[:space:]] ]] && unset "MODULES[i]"
+    [[ " ${EXCLUDE_MODULES[*]} " =~ [[:space:]]${MODULES[i]}[[:space:]] ]] && unset "MODULES[$i]"
   done
     
   # Sanitize MODULES
   for i in "${!MODULES[@]}"; do
     if ! [[ " ${ALL_MODULES[*]} " =~ [[:space:]]${MODULES[i]}[[:space:]] ]]; then
       echo "Module ${MODULES[$i]} not in the module list"
-      echo "Available modules:"
-      printf "%s\n" "${ALL_MODULES[@]}"
       read -r -p "Enter replacement module name: " module
       ! [[ " ${ALL_MODULES[*]} " =~ [[:space:]]${module}[[:space:]] ]] || (echo "RTFM"; return 1)
       MODULES[i]="$module"
@@ -2376,7 +2365,7 @@ main() {
   for i in "${!WRAPPERS[@]}"; do
     if ! [[ " ${ALL_WRAPPERS[*]} " =~ [[:space:]]${WRAPPERS[i]}[[:space:]] ]]; then
       echo "Wrapper ${WRAPPERS[$i]} not in the module list, removing"
-      unset "WRAPPERS[i]" "WRAPPERS[$((i+1))]"
+      unset "WRAPPERS[$i]" "WRAPPERS[$((i+1))]"
     fi
     continue 2 # skip the arguments string
   done
@@ -2426,12 +2415,13 @@ main() {
   unset MODULES WRAPPERS EXCLUDE_MODULES STUDIES SET_STUDIES YES
 }
 
-# (Safe) main loop
-if main "$@"; then
-  for ((i=1; i<2; i++)); do
-    main &&
-    i=0
-  done
-fi
+# (Safe) main loop if we want to rerun automatically
+# if main "$@"; then
+#   for ((i=1; i<2; i++)); do
+#     main &&
+#     i=0
+#   done
+# fi
 
+main "$@"
 exit $?