Important updates (Closes #1)

process_by_sampleid.sh

 #!/bin/bash
 
-set -e
-trap 'last_command=$current_command; current_command=$BASH_COMMAND' DEBUG
-trap 'if [ "$?" -ne "0" ]; then echo "\"${last_command}\" command filed with exit code different from 0"; fi' EXIT
+# set -e
+# trap 'last_command=$current_command; current_command=$BASH_COMMAND' DEBUG
+# trap 'if [ "$?" -ne "0" ]; then echo "\"${last_command}\" command filed with exit code different from 0"; fi' EXIT
 
-if [[ $# -ne 3 ]]; then
+###### FUNCTIONS
+
+usage()
+{
     echo ""
     echo "USAGE:"
-    echo "process_by_sampleid.sh <SAMPLE_ID> <STORAGE_ACCT_FASTQ>/<FILESHARE_FASTQ> <STORAGE_ACCT_RESULTS>/<FILESHARE_RESULTS_PREFIX>"
+    echo "process_by_sampleid.sh [OPTIONS] -s <SAMPLE_ID> -i <STORAGE_ACCT_FASTQ>/<FILESHARE_FASTQ> -o <STORAGE_ACCT_RESULTS>/<FILESHARE_RESULTS_PREFIX>"
     echo "  - SAMPLE_ID: e.g. SL284409"
     echo "  - STORAGE_ACCT_FASTQ: e.g. forneurosysmedutvcompute"
     echo "  - FILESHARE_FASTQ: Name of the fileshare, e.g. wgbsfqdata"
     echo "  - STORAGE_ACCT_RESULTS: e.g. forneurosysmedutvcompute"
     echo ""
+    echo " EXTRA OPTIONS"
+    echo "   -f --force"
+    echo "          Overwrite previous results, if any."
+    echo "   -m --monitor"
+    echo "          Monitor RAM and CPU usage."
+    echo "   -u --subset INTEGER"
+    echo "          Subset reads to test pipeline (to INTEGER)."
+    echo "   --strategies STRATEGIES"
+    echo "          Only run selected strategies. Possible choices of STRATEGIES: "1", "2", "3""
+    echo "          or any combination of them, comma-separated (e.g. "1,2,3", "2,3")."
+    echo "   --input-samba-credentials FILE"
+    echo "          File with the samba credentials to mount input fileshare (fastqs)."
+    echo "   --output-samba-credentials FILE"
+    echo "          File with the samba credentials to mount output fileshare (results)."
     echo "There should be files in the /etc/smbcredentials/ folder named <STORAGE_ACCT_FASTQ>.cred and <STORAGE_ACCT_RESULTS>.cred (the storage account can be the same for both input fastqs and results, then only one file is needed). This/these file/s contain the username=... and the password=... lines with the storage account name and the key."
     echo ""
     echo "For the results, a new fileshare will be created for the sample analysed, whose name will be a concatenation of the form <FILESHARE_RESULTS_PREFIX>-<SAMPLE_ID>. If the fileshare name exists and is not empty, the program will fail."
     echo ""
     echo "EXAMPLE: process_by_sampleid.sh SL284405 forneurosysmedutvcompute/wgbsfqdata forneurosysmedutvcompute/prefix"
-    exit 1
+}
+
+###### PARSE CMDLINE
+
+sample_id=
+storageAccountFastq=
+shareNameFastq=
+storageAccountResults=
+shareNameResults=
+FORCE=0
+MONITOR=0
+SUBSET=0
+STRATEGIES="1,2,3"
+
+smbCredFastq=
+smbCredResults=
+
+while [ "$1" != "" ]; do
+    case $1 in
+        -s | --sample-id )               shift
+                                         sample_id=$1
+                                         ;;
+        -i | --input-storage-account )   shift
+                                         storageAccountFastq=${1%/*}
+                                         shareNameFastq=${1#*/}
+                                         ;;
+        -o | --output-storage-account )  shift
+                                         storageAccountResults=${1%/*}
+                                         shareNameResults=${1#*/}
+                                         ;;
+        --input-samba-credentials )      shift
+                                         smbCredFastq=$1
+                                         ;;
+        --output-samba-credentials )     shift
+                                         smbCredResults=$1
+                                         ;;
+        -f | --force )                   FORCE=1
+                                         ;;
+        -m | --monitor )                 MONITOR=1
+                                         ;;
+        -u | --subset )                  shift
+                                         SUBSET=$1
+                                         ;;
+        --strategies )                   shift
+                                         STRATEGIES=$1
+                                         ;;
+        -h | --help )                    usage
+                                         exit
+                                         ;;
+        * )                              usage
+                                         exit 1
+    esac
+    shift
+done
+
+if [[ "$sample_id" == "" ]]; then
+    echo "[ERROR] -s option is mandatory."
+    exit 2
+fi
+if [[ "$storageAccountFastq" == "" ]]; then
+    echo "[ERROR] -i option is mandatory, e.g. forneurosysmedutvcompute/wgbsfqdata"
+    exit 2
+fi
+if [[ "$shareNameFastq" == "" ]]; then
+    echo "[ERROR] -i option is mandatory, e.g. forneurosysmedutvcompute/wgbsfqdata"
+    exit 2
+fi
+if [[ "$storageAccountResults" == "" ]]; then
+    echo "[ERROR] -o option is mandatory, e.g. forneurosysmedutvcompute/prefix"
+    exit 2
+fi
+if [[ "$shareNameResults" == "" ]]; then
+    echo "[ERROR] -o option is mandatory, e.g. forneurosysmedutvcompute/prefix"
+    exit 2
+fi
+
+if [[ "$SUBSET" == "" ]]; then
+    echo "[ERROR] -u option needs an integer, e.g. -u 10000"
+    exit 2
+fi
+re='^[0-9]+$'
+if ! [[ $SUBSET =~ $re ]]; then
+       echo "[ERROR] -u option needs and integer, e.g. -u 100000"
+       exit 2
 fi
 
-sample_id=$1
-storageAccountFastq=${2%/*}
-shareNameFastq=${2#*/}
-storageAccountResults=${3%/*}
-shareNameResults=${3#*/}-${sample_id}
-shareNameResults=$( echo ${shareNameResults} | tr '[:upper:]' '[:lower:]' )
+shareNameResults=$( echo ${shareNameResults}-${sample_id} | tr '[:upper:]' '[:lower:]' )
+
+if [[ "$smbCredFastq" == "" ]]; then
+    smbCredFastq=/etc/smbcredentials/${storageAccountFastq}.cred
+    echo "[WARNING] Using default samba credentials file path: $smbCredFastq"
+fi
+if [[ "$smbCredResults" == "" ]]; then
+    smbCredResults=/etc/smbcredentials/${storageAccountResults}.cred
+    echo "[WARNING] Using default samba credentials file path: $smbCredResults"
+fi
 
-smbCredFastq=/etc/smbcredentials/${storageAccountFastq}.cred
-smbCredResults=/etc/smbcredentials/${storageAccountResults}.cred
 
 echo
 echo "INPUT / VARIABLES"
@@ -41,6 +143,10 @@ echo "- storageAccountResults: ${storageAccountResults}"
 echo "- shareNameResults: ${shareNameResults}"
 echo "- smbCredFastq: ${smbCredFastq}"
 echo "- smbCredResults: ${smbCredResults}"
+echo "- FORCE: $FORCE"
+echo "- MONITOR: $MONITOR"
+echo "- SUBSET: $SUBSET"
+echo "- STRATEGIES: ${STRATEGIES}"
 echo
 
 echo "ACTUALLY DOING STUFF"
@@ -66,6 +172,7 @@ fi
 
 echo "   [DONE]"
 
+
 echo ">> Preparing input file share... "
 # Check that fastq file share exists
 exists=$( az storage share exists --account-name ${storageAccountFastq} --account-key ${storageAccountKeyFastq} --name ${shareNameFastq} --output tsv )
@@ -102,14 +209,9 @@ echo "  *Input FASTQ file share mounted in \"${fastq_folder}\""
 echo ">> Preparing output file share... "
 # Check that results fileshare does NOT exist or issue warning and wait
 exists=$( az storage share exists --account-name ${storageAccountResults} --account-key ${storageAccountKeyResults} --name "${shareNameResults}" --output tsv )
-if [[ $exists == "True" ]]; then
-    echo "   [WARNING] File share \"${shareNameResults}\" in \"${storageAccountResults}\" ALREADY EXISTS!"
-    echo -n "   OVERWRITING FILES IN 30 SECONDS..."
-    for i in {30..01}; do
-        echo -ne "\b\b\b\b\b\b\b\b\b\b\b\b\b${i} SECONDS..."
-        sleep 1
-    done
-    echo -e "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b!                "
+if [[ $exists == "True" && $FORCE == 0 ]]; then
+    echo "   [ERROR] File share \"${shareNameResults}\" in \"${storageAccountResults}\" ALREADY EXISTS!"
+    echo "           To overwrite, use -f option"
 fi
 
 results_folder="/fileshares/${storageAccountResults}/${shareNameResults}"
@@ -141,10 +243,13 @@ sudo mount -t cifs //${storageAccountResults}.file.core.windows.net/${shareNameR
 if [[ $? != 0 ]]; then
     echo "   [ERROR] Could not mount file share for RESULTS"
     exit 2
+else
+    cd ${results_folder}
 fi
+
 echo "   [DONE]"
 
-    echo "   *Results file share mounted in \"${results_folder}\""
+echo "   *Results file share mounted in \"${results_folder}\""
 
 
 
@@ -159,7 +264,7 @@ if [[ $num_files -eq 0 ]]; then
 fi
 
 echo "   [DONE]"
-echo "  *Number of FASTQ files retrieved: ${num_files}"
+echo "  *Number of FASTQ files found: ${num_files}"
 
 declare -a R1
 declare -a R2
@@ -178,15 +283,28 @@ done
 
 
 # RUNNIN' THE PIPELINE
+opts=" -o ${results_folder} "
+if [[ $FORCE != 0 ]]; then
+    opts="${opts}-f "
+fi
+if [[ $MONITOR != 0 ]]; then
+    opts="${opts}-m "
+fi
+if [[ $SUBSET != 0 ]]; then
+    opts="${opts}-u ${SUBSET} "
+fi
+opts="${opts}-s $STRATEGIES "
+
+PIPELINE_CMD="bash /home/gsnido/wgbs-pipeline/wgbs-pipeline.sh${opts}${R1[*]} ${R2[*]}"
 echo ">> RUNNING THE PIPELINE..."
-echo
-bash /home/gsnido/wgbs-pipeline/wgbs-pipeline.sh ${R1[*]} ${R2[*]} ${results_folder}
+echo "   CMD = \"${PIPELINE_CMD}\""
+${PIPELINE_CMD}
 echo
 echo "**************FINISHED**************"
 echo
 
 # Get space used by files in share in GB
-sleep 60
+sleep 30
 size1=$( az storage share stats --name ${shareNameResults} --account-name ${storageAccountResults} --account-key ${storageAccountKeyResults} )
 size2=$( du -sh -BG ${results_folder} | cut -f 1 | tr -d 'G' )
 size=$(( size1 > size2 ? size1 + 10 : size2 + 10))
@@ -197,6 +315,11 @@ echo " Resizing fileshare quota to $size..."
 az storage share update --name ${shareNameResults} --account-name ${storageAccountResults} --account-key ${storageAccountKeyResults} --quota ${size}
 
 echo 
+cd ~
+echo " Unmounting FASTQ folder..."
+sudo umount ${fastq_folder}
+echo " Unmounting results folder..."
+sudo umount ${results_folder}
 echo "   [DONE]"