Finish draft pipeline scripts for TEannot

TEannot/TEannot_AllSteps.sh

@@ -67,28 +67,29 @@ jid_step4=$(sbatch \
     TEannot_Step4.sh | \
     cut -d" " -f4)
 
-#jid_step5=$(sbatch \
-#    --kill-on-invalid-dep=yes \
-#    --dependency=afterok:$jid_step4 \
-#    TEannot_Step5.sh | \
-#    cut -d" " -f4)
-
-#jid_step6=$(sbatch \
-#    --kill-on-invalid-dep=yes \
-#    --dependency=afterok:$jid_step1 \
-#    TEannot_Step6.sh | \
-#    cut -d" " -f4)
-
-#jid_step7=$(sbatch \
-#    --kill-on-invalid-dep=yes \
-#    --dependency=afterok:$jid_step3:$jid_step5:$jid_step6 \
-#    TEannot_Step7.sh | \
-#    cut -d" " -f4)
-
-#jid_step8=$(sbatch \
-#    --kill-on-invalid-dep=yes \
-#    --dependency=afterok:$jid_step7 \
-#    TEannot_Step8.sh | \
-#    cut -d" " -f4)
+jid_step5=$(sbatch \
+    --kill-on-invalid-dep=yes \
+    --dependency=afterok:$jid_step4 \
+    TEannot_Step5.sh | \
+    cut -d" " -f4)
+
+jid_step6=$(sbatch \
+    --kill-on-invalid-dep=yes \
+    --dependency=afterok:$jid_step1 \
+    --array=0-1 \
+    TEannot_Step6.sh | \
+    cut -d" " -f4)
+
+jid_step7=$(sbatch \
+    --kill-on-invalid-dep=yes \
+    --dependency=afterok:$jid_step3:$jid_step5:$jid_step6 \
+    TEannot_Step7.sh | \
+    cut -d" " -f4)
+
+jid_step8=$(sbatch \
+    --kill-on-invalid-dep=yes \
+    --dependency=afterok:$jid_step7 \
+    TEannot_Step8.sh | \
+    cut -d" " -f4)
 
 echo "Finished submitting all jobs at $(date)"

TEannot/TEannot_Step5.sh

@@ -1,13 +1,13 @@
 #!/bin/bash
 
 #SBATCH --nodes=1
-#SBATCH --ntasks=4
+#SBATCH --ntasks=2
 #SBATCH --mem-per-cpu=2G
 #SBATCH --time=1-00:00:00
-#SBATCH --output=TEAnnot_step5.stdout
+#SBATCH --output=TEannot-step5.stdout
 #SBATCH --job-name="S5_TEannot"
 #SBATCH -p intel
 
 module load repet/2.5
-source config.txt
+
 TEannot.py -P $ProjectName -C TEannot.cfg -S 5

TEannot/TEannot_Step6.sh

@@ -4,11 +4,43 @@
 #SBATCH --ntasks=4
 #SBATCH --mem-per-cpu=2G
 #SBATCH --time=1-00:00:00
-#SBATCH --output=TEAnnot_step6.stdout
+#SBATCH --output=TEannot-step6-%a.stdout
 #SBATCH --job-name="S6_TEannot"
 #SBATCH -p intel
 
 module load repet/2.5
-source config.txt
-TEannot.py -P $ProjectName -C TEannot.cfg -S 6 -b tblastx
-TEannot.py -P $ProjectName -C TEannot.cfg -S 6 -b blastx
+
+# REPET TEannot - Step 6
+# Align RepBase databanks to genome
+
+LOCAL_ALIGNERS=("blastx" "tblastx")
+LOCAL_ALIGNER_ABRS=("x" "tx")
+LOCAL_ALIGNER=${LOCAL_ALIGNERS[$SLURM_ARRAY_TASK_ID]}
+LCL_ALN=${LOCAL_ALIGNER_ABRS[$SLURM_ARRAY_TASK_ID]}
+
+if [ $SLURM_ARRAY_TASK_ID -eq '0' ]; then
+    OUT_DIR="${ProjectName}_TEdetect/bankBLRx"
+elif [ $SLURM_ARRAY_TASK_ID -eq '1' ]; then
+    OUT_DIR="${ProjectName}_TEdetect/bankBLRtx"
+else
+    echo "SLURM array improperly set up"
+    exit 1
+fi
+
+if [ ! -d "${ProjectName}_TEdetect/bankBLR${LCL_ALN}" ]; 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 ${ProjectName}_chk_bankBLR${LCL_ALN}_path;" \
+    "DROP TABLE IF EXISTS ${ProjectName}_chr_bankBLR${LCL_ALN}_path;" \
+    "DROP TABLE IF EXISTS ${ProjectName}_bankBLR${LCL_ALN}_nt_seq;" \
+    "DROP TABLE IF EXISTS ${ProjectName}_bankBLR${LCL_ALN}_prot_seq;" | \
+    mysql -h $MYSQL_HOST -u$MYSQL_USER -p$MYSQL_PASS $MYSQL_DB
+
+    TEannot.py -P $ProjectName -C TEannot.cfg -S 6 -b $LOCAL_ALIGNER
+else
+    echo "Step 6 output folder detected, skipping..."
+fi

TEannot/TEannot_Step7.sh

@@ -4,10 +4,10 @@
 #SBATCH --ntasks=1
 #SBATCH --mem-per-cpu=2G
 #SBATCH --time=1-00:00:00
-#SBATCH --output=TEAnnot_step7.stdout
+#SBATCH --output=TEannot-step7.stdout
 #SBATCH --job-name="S7_TEannot"
 #SBATCH -p intel
 
 module load repet/2.5
-source config.txt
+
 TEannot.py -P $ProjectName -C TEannot.cfg -S 7

TEannot/TEannot_Step8.sh

@@ -4,10 +4,13 @@
 #SBATCH --ntasks=1
 #SBATCH --mem-per-cpu=2G
 #SBATCH --time=1-00:00:00
-#SBATCH --output=TEAnnot_step8.stdout
+#SBATCH --output=TEannot-step8.stdout
 #SBATCH --job-name="S8_TEannot"
 #SBATCH -p intel
 
 module load repet/2.5
-source config.txt
-TEannot.py -P $ProjectName -C TEannot.cfg -S 8 -o GFF3
+
+if [ ! -d "${ProjectName}_GFF3" ]; then
+    TEannot.py -P $ProjectName -C TEannot.cfg -S 8 -o GFF3
+fi
+