Merge branch 'dev' of ssh://git-r3lab-server.uni.lu:8022/shaman.narayanasamy/IMP into dev

IMP

@@ -91,11 +91,14 @@ if __name__ == '__main__':
 
     # CL
     cmd = ['docker', 'run'] + mount_points + envs
-    # reconfigure entrypoint if --enter flag is specified
+    # if --enter flag is specified, add attach tty and set mode to interactive
     if args['--enter']:
-        cmd += ['--entrypoint /bin/bash -it']
+        cmd += ['-it']
     # add container name and commands to pass to snakemake
     cmd += [container_name] + args['COMMANDS']
+    # if --enter flag is specified, change the command
+    if args['--enter']:
+        cmd += ['/bin/bash']
     # parse CL correctly
     cmd = shlex.split(' '.join(cmd))
     print("Executing", '"', ' '.join(cmd), '"')

docker/Dockerfile

@@ -216,5 +216,4 @@ ENV PATH /home/imp/lib/trinityrnaseq_r20140717:/home/imp/lib/Platypus_0.8.1:/hom
 
 VOLUME ["/data", "/output"]
 WORKDIR /home/imp/integrated-metaomic-pipeline
-ENTRYPOINT ["snakemake"]
-CMD ["ALL"]
+CMD ["snakemake", "ALL"]

rules/Assembly/MGMT.rules

@@ -21,13 +21,16 @@ rule ASSEMBLY_MT_MEGAHIT_1:
         outdir = "{dir}/MT.assembly_1".format(dir=A_OUT)
     shell:
         """
+        if [ -d "{params.outdir}" ]; then
+            rm -rf {params.outdir}
+        fi
         MAX_READ_LEN=$(cat {input} | sed -n '1~4s/^@/>/p;2~4p' |\
         awk '$0 ~ \">\" {{c=0\"\t\"; }} $0 !~ \">\" {{c+=length($0); max=(max>c)?max:c;}} END {{print max}}')
         echo "Max read length: $MAX_READ_LEN"
         MEMBYTES=$(({MEMTOTAL}*1000*1000*1000))
         echo "Available memory in bytes: $MEMBYTES"
         megahit -o {params.outdir} --cpu-only -m $MEMBYTES --mem-flag 1\
-        -l $MAX_READ_LEN --input-cmd "cat {input}" -t {THREADS} --continue >> {log} 2>&1
+        -l $MAX_READ_LEN --input-cmd "cat {input}" -t {THREADS} >> {log} 2>&1
         """
 
 
@@ -80,13 +83,17 @@ rule ASSEMBLY_MT_MEGAHIT_2:
         outdir = "{dir}/MT.assembly_2".format(dir=A_OUT)
     shell:
         """
+        if [ -d "{params.outdir}" ]; then
+            rm -rf {params.outdir}
+        fi
+
         MAX_READ_LEN=$(cat {input} | sed -n '1~4s/^@/>/p;2~4p' | \
         awk '$0 ~ \">\" {{c=0\"\t\"; }} $0 !~ \">\" {{c+=length($0); max=(max>c)?max:c;}} END {{print max}}')
         echo "Max read length: $MAX_READ_LEN"
         MEMBYTES=$(({MEMTOTAL}*1000*1000*1000))
         echo "Available memory in bytes: $MEMBYTES"
         megahit -o {params.outdir} --cpu-only -m $MEMBYTES --mem-flag 1 -l $MAX_READ_LEN \
-        --input-cmd "cat {input}" -t {THREADS} --continue >> {log} 2>&1
+        --input-cmd "cat {input}" -t {THREADS} >> {log} 2>&1
         """
 
 rule ASSEMBLY_MT_CAT_MEGAHIT:
@@ -424,13 +431,16 @@ rule ASSEMBLY_MGMT_FAST_1:
         outdir = "{dir}/MGMT.assembly_1".format(dir=A_OUT)
     shell:
         """
+        if [ -d "{params.outdir}" ]; then
+            rm -rf {params.outdir}
+        fi
         MAX_READ_LEN=$(cat {input} | sed -n '1~4s/^@/>/p;2~4p' |\
         awk '$0 ~ \">\" {{c=0\"\t\"; }} $0 !~ \">\" {{c+=length($0); max=(max>c)?max:c;}} END {{print max}}')
         echo "Max read length: $MAX_READ_LEN"
         MEMBYTES=$(({MEMTOTAL}*1000*1000*1000))
         echo "Available memory in bytes: $MEMBYTES"
         megahit -o {params.outdir} --cpu-only -m $MEMBYTES --mem-flag 1\
-        -l $MAX_READ_LEN --input-cmd "cat {input}" -t {THREADS} --continue >> {log} 2>&1
+        -l $MAX_READ_LEN --input-cmd "cat {input}" -t {THREADS} >> {log} 2>&1
         """
 
 rule ASSEMBLY_MGMT_EXTRACT_UNMAPPED_FROM_FAST_1:
@@ -482,9 +492,12 @@ rule ASSEMBLY_MGMT_FAST_2:
         outdir = "{dir}/MGMT.assembly_2".format(dir=A_OUT)
     shell:
         """
+        if [ -d "{params.outdir}" ]; then
+            rm -rf {params.outdir}
+        fi
         MAX_READ_LEN=$(cat {input} | sed -n '1~4s/^@/>/p;2~4p' | \
         awk '$0 ~ \">\" {{c=0\"\t\"; }} $0 !~ \">\" {{c+=length($0); max=(max>c)?max:c;}} END {{print max}}')
         MEMBYTES=$(({MEMTOTAL}*1000*1000*1000))
         megahit -o {params.outdir} --cpu-only -m $MEMBYTES --mem-flag 1 -l $MAX_READ_LEN \
-        --input-cmd "cat {input}" -t {THREADS} --continue >> {log} 2>&1
+        --input-cmd "cat {input}" -t {THREADS} >> {log} 2>&1
         """

src/config.imp.json

@@ -61,7 +61,7 @@
         ]
     },
     "prokka": {
-        "pkg_url": "http://www.vicbioinformatics.com/prokka-1.11.tar.gz",
+        "pkg_url": "https://webdav-r3lab.uni.lu/public/R3lab/IMP/prokka-1.11.tar.gz",
         "databases": [
             "cm/Bacteria.i1i",
             "genus/Staphylococcus.phr",