From 8bf2106f7fc27649e3c96d1d5cb468ed80da3ede Mon Sep 17 00:00:00 2001
From: Travis Wrightsman <twrightsman@users.noreply.github.com>
Date: Fri, 7 Jul 2017 16:52:04 -0700
Subject: [PATCH] Improve TEannot steps 2, 3, and 4 job scripts

---
 TEannot/TEannot_AllSteps.sh | 27 +++++++++++++++++----------
 TEannot/TEannot_Step2.sh    |  2 ++
 TEannot/TEannot_Step3.sh    | 31 +++++++++++++++++++++++++++----
 TEannot/TEannot_Step4.sh    | 18 ++++++++++++------
 4 files changed, 58 insertions(+), 20 deletions(-)

diff --git a/TEannot/TEannot_AllSteps.sh b/TEannot/TEannot_AllSteps.sh
index 296aa8f..c838816 100644
--- a/TEannot/TEannot_AllSteps.sh
+++ b/TEannot/TEannot_AllSteps.sh
@@ -15,6 +15,7 @@ export ProjectName=$(grep "project_name" TEannot.cfg | cut -d" " -f2)
 # (!) modify these to your project/environment
 ## (only choose what REPET supports)
 export ALIGNERS_AVAIL="BLR+RM+CEN"
+export SSR_DETECTORS_AVAIL="TRF+RMSSR"
 
 # ALIGNERS_AVAIL has to be a string because bash arrays cannot be passed
 # directly to SLURM jobs; so the string is split into an array here and
@@ -23,6 +24,9 @@ IFS='+' read -ra ALIGNERS_AVAIL_ARRAY <<< "$ALIGNERS_AVAIL"
 # ${#ALIGNERS_AVAIL_ARRAY[@]} gives length of ALIGNERS_AVAIL_ARRAY array
 NUM_ALIGNERS=${#ALIGNERS_AVAIL_ARRAY[@]}
 
+IFS='+' read -ra SSR_DETECTORS_AVAIL_ARRAY <<< "$SSR_DETECTORS_AVAIL"
+NUM_SSR_DETECTORS=${#SSR_DETECTORS_AVAIL_ARRAY[@]}
+
 # Clear the jobs table for the current project
 ## in case last run failed for some reason while sub-jobs were running
 MYSQL_HOST=$(grep "repet_host" TEannot.cfg | cut -d" " -f2)
@@ -48,17 +52,20 @@ jid_step2=$(sbatch \
     TEannot_Step2.sh | \
     cut -d" " -f4)
 
-#jid_step3=$(sbatch \
-#    --kill-on-invalid-dep=yes \
-#    --dependency=afterok:$jid_step2 \
-#    TEannot_Step3.sh | \
-#    cut -d" " -f4)
+jid_step3=$(sbatch \
+    --export=ProjectName,ALIGNERS_AVAIL \
+    --kill-on-invalid-dep=yes \
+    --dependency=afterok:$jid_step2 \
+    TEannot_Step3.sh | \
+    cut -d" " -f4)
 
-#jid_step4=$(sbatch \
-#    --kill-on-invalid-dep=yes \
-#    --dependency=afterok:$jid_step1 \
-#    TEannot_Step4.sh | \
-#    cut -d" " -f4)
+jid_step4=$(sbatch \
+    --export=ProjectName,SSR_DETECTORS_AVAIL \
+    --kill-on-invalid-dep=yes \
+    --dependency=afterok:$jid_step1 \
+    --array=0-$(( $NUM_SSR_DETECTORS - 1 )) \
+    TEannot_Step4.sh | \
+    cut -d" " -f4)
 
 #jid_step5=$(sbatch \
 #    --kill-on-invalid-dep=yes \
diff --git a/TEannot/TEannot_Step2.sh b/TEannot/TEannot_Step2.sh
index 30a8267..407f78a 100644
--- a/TEannot/TEannot_Step2.sh
+++ b/TEannot/TEannot_Step2.sh
@@ -38,4 +38,6 @@ ALIGNER=${ALIGNERS_AVAIL_ARRAY[$(( $SLURM_ARRAY_TASK_ID % $NUM_ALIGNERS ))]}
 
 if [ ! -d "${OUT_DIR}/${ALIGNER}" ]; then
     TEannot.py -P $ProjectName -C TEannot.cfg -S 2 -a $ALIGNER $CMD_SUFFIX
+else
+    echo "Step 2 output folder detected, skipping..."
 fi
diff --git a/TEannot/TEannot_Step3.sh b/TEannot/TEannot_Step3.sh
index 44f6061..587dc81 100644
--- a/TEannot/TEannot_Step3.sh
+++ b/TEannot/TEannot_Step3.sh
@@ -4,12 +4,35 @@
 #SBATCH --ntasks=1
 #SBATCH --mem-per-cpu=2G
 #SBATCH --time=1-00:00:00
-#SBATCH --output=TEAnnot_step3.stdout
+#SBATCH --output=TEannot-step3.stdout
 #SBATCH --job-name="S3_TEannot"
 #SBATCH -p intel
 
 module load repet/2.5
-source config.txt
-# REPET - Step 3
 
-TEannot.py -P $ProjectName -C TEannot.cfg -S 3 -c BLR+RM+CEN
+# REPET TEannot - Step 3
+# Filter and combine HSPs from Step 2 alignment
+
+if  [ ! -n "$ProjectName" ] || [ ! -n "$ALIGNERS_AVAIL" ]; then
+    echo 'One or more environment variables required by this script' \
+    'are unset. Either run this script through the scheduler script or' \
+    'set the variable(s) and use the --export option of sbatch before' \
+    'restarting.'
+    exit 1
+fi
+
+if [ ! -d "${ProjectName}_TEdetect/Comb" ]; then
+    # if re-running step, drop MySQL tables
+    MYSQL_HOST=$(grep "repet_host" TEannot.cfg | cut -d" " -f2)
+    MYSQL_USER=$(grep "repet_user" TEannot.cfg | cut -d" " -f2)
+    MYSQL_PASS=$(grep "repet_pw" TEannot.cfg | cut -d" " -f2)
+    MYSQL_DB=$(grep "repet_db" TEannot.cfg | cut -d" " -f2)
+
+    echo "DROP TABLE IF EXISTS ${ProjectName}_chk_allTEs_path;" \
+    "DROP TABLE IF EXISTS ${ProjectName}_chr_allTEs_path;" | \
+    mysql -h $MYSQL_HOST -u$MYSQL_USER -p$MYSQL_PASS $MYSQL_DB
+
+    TEannot.py -P $ProjectName -C TEannot.cfg -S 3 -c $ALIGNERS_AVAIL
+else
+    echo "Step 3 output folder detected, skipping..."
+fi
diff --git a/TEannot/TEannot_Step4.sh b/TEannot/TEannot_Step4.sh
index 53ae3dc..21523c4 100644
--- a/TEannot/TEannot_Step4.sh
+++ b/TEannot/TEannot_Step4.sh
@@ -4,14 +4,20 @@
 #SBATCH --ntasks=1
 #SBATCH --mem-per-cpu=2G
 #SBATCH --time=1-00:00:00
-#SBATCH --output=TEAnnot_step4.stdout
+#SBATCH --output=TEannot-step4-%a.stdout
 #SBATCH --job-name="S4_TEannot"
 #SBATCH -p intel
 
 module load repet/2.5
-source config.txt
-# REPET - Step 4
 
-TEannot.py -P $ProjectName -C TEannot.cfg -S 4 -s TRF
-TEannot.py -P $ProjectName -C TEannot.cfg -S 4 -s Mreps
-TEannot.py -P $ProjectName -C TEannot.cfg -S 4 -s RMSSR
+# REPET TEannot - Step 4
+# Search for satellites in the genomic sequence
+
+IFS='+' read -ra SSR_DETECTORS_AVAIL_ARRAY <<< "${SSR_DETECTORS_AVAIL}"
+SSR_DETECTOR=${SSR_DETECTORS_AVAIL_ARRAY[$SLURM_ARRAY_TASK_ID]}
+
+if [ ! -d "${ProjectName}_SSRdetect/${SSR_DETECTOR}" ]; then
+    TEannot.py -P $ProjectName -C TEannot.cfg -S 4 -s $SSR_DETECTOR
+else
+    echo "Step 4 output folder detected, skipping..."
+fi