From 61d977b241140b1c09e74c09b1cca55f632f3f11 Mon Sep 17 00:00:00 2001
From: Shaman <shaman.narayanasamy@uni.lu>
Date: Fri, 27 Mar 2015 17:36:46 +0100
Subject: [PATCH] Add Trinity MT assembly module. Commented out at the moment

---
 rules/Assembly/MGMT.rules   | 121 +++++++++++++++++++++++++++++++++---
 rules/Assembly/master.rules |   6 +-
 2 files changed, 117 insertions(+), 10 deletions(-)

diff --git a/rules/Assembly/MGMT.rules b/rules/Assembly/MGMT.rules
index 26d2c21..58c5633 100644
--- a/rules/Assembly/MGMT.rules
+++ b/rules/Assembly/MGMT.rules
@@ -1,8 +1,12 @@
-################
-# ## MT SECTION
-################
+###################################################################################################
+# MT ASSEEMBLY SECTION
+###################################################################################################
 
-rule ASSEMBLY_MT_ASSEMBLY_1:
+#########################
+## ## MT MEGAHIT ASSEMBLY
+#########################
+
+rule ASSEMBLY_MT_MEGAHIT_1:
     log:
         A_LOG
     benchmark:
@@ -27,7 +31,7 @@ rule ASSEMBLY_MT_ASSEMBLY_1:
         """
 
 
-rule ASSEMBLY_MT_EXTRACT_UNMAPPED_FROM_ASSEMBLY_1:
+rule ASSEMBLY_MT_EXTRACT_UNMAPPED_FROM_MEGAHIT_1:
     log:
         A_LOG
     benchmark:
@@ -60,7 +64,7 @@ rule ASSEMBLY_MT_EXTRACT_UNMAPPED_FROM_ASSEMBLY_1:
         rm -rf $BUFFER* $TMP_FILE
         """
 
-rule ASSEMBLY_MT_ASSEMBLY_2:
+rule ASSEMBLY_MT_MEGAHIT_2:
     log:
         A_LOG
     benchmark:
@@ -83,7 +87,7 @@ rule ASSEMBLY_MT_ASSEMBLY_2:
         --input-cmd "cat {input}" -t {THREADS} --continue > {log} 2>&1
         """
 
-rule ASSEMBLY_MT_CAT_ASSEMBLY:
+rule ASSEMBLY_MT_CAT_MEGAHIT:
     log:
         A_LOG
     benchmark:
@@ -97,6 +101,109 @@ rule ASSEMBLY_MT_CAT_ASSEMBLY:
         """
         cat {input[0]} {input[1]} | awk '/^>/{{print ">contig_MT_" ++i; next}}{{print}}' > {output}
         """
+
+#########################
+## ## MT MEGAHIT ASSEMBLY
+#########################
+#
+#rule ASSEMBLY_MT_ASSEMBLY_TRINITY_1:
+#    log:
+#        A_LOG
+#    benchmark:
+#        "%s/benchmarks/ASSEMBLY_MT_ASSEMBLY_1.json" % A_OUT
+#    input:
+#        preprocessed_mt('R1'),
+#        preprocessed_mt('R2'),
+#        preprocessed_mt('SE')
+#    output:
+#        '{dir}/MT.assembly_1/final.contigs.fa'.format(dir=A_OUT)
+#    params:
+#        outdir = "{dir}/MT.assembly_1".format(dir=A_OUT)
+#    shell:
+#        """
+#        Trinity --seqType fq\
+#        --left {input[0]} --right {input[1]} --single {input[2]}\
+#        --output {output[0]}\
+#        --JM {MEMTOTAL}G --CPU {THREADS}\
+#        --inchworm_cpu {THREADS} --bflyHeapSpaceMax {MEMTOTAL}G\
+#        --bflyCPU {THREADS} --bflyCalculateCPU\
+#        --normalize_max_read_cov 25  --full_cleanup\
+#        """
+#
+#
+#rule ASSEMBLY_MT_EXTRACT_UNMAPPED_FROM_TRINITY_1:
+#    log:
+#        A_LOG
+#    benchmark:
+#        "%s/benchmarks/ASSEMBLY_MT_EXTRACT_UNMAPPED_FROM_ASSEMBLY_1.json" % A_OUT
+#    input:
+#        preprocessed_mt('R1'),
+#        preprocessed_mt('R2'),
+#        preprocessed_mt('SE'),
+#        '{dir}/MT.assembly_1/final.contigs.fa'.format(dir=A_OUT),
+#        expand('{dir}/MT.assembly_1/final.contigs.fa.{ext}', dir=A_OUT, ext=['amb', 'bwt', 'pac', 'sa', 'ann'])
+#    output:
+#        expand('{dir}/{name}', name=[
+#            'MT.R1.unmapped.fq',
+#            'MT.R2.unmapped.fq',
+#            'MT.SE.unmapped.fq'], dir=A_OUT)
+#    shell:
+#        """
+#        TMP_FILE=$(mktemp --tmpdir={TMPDIR} -t "alignment_XXXXXX.bam")
+#        BUFFER=$(mktemp --tmpdir={TMPDIR} -t "alignment_buffer_XXXXXX.bam")
+#        bwa mem -v 1 -t {THREADS} {input[3]} {input[0]} {input[1]} | samtools view -@ {THREADS} -bS - > $TMP_FILE
+#        samtools merge -@ {THREADS} -u - \
+#        <(samtools view -@ {THREADS} -u  -f 4 -F 264 $TMP_FILE) \
+#        <(samtools view -@ {THREADS} -u -f 8 -F 260 $TMP_FILE) \
+#        <(samtools view -@ {THREADS} -u -f 12 -F 256 $TMP_FILE) | \
+#        samtools view -@ {THREADS} -bF 0x800 -  | samtools sort -o -@ {THREADS} -m {MEMCORE}G -n - $BUFFER | \
+#        bamToFastq -i stdin -fq {output[0]} -fq2 {output[1]}
+#        bwa mem -v 1 -t {THREADS} {input[3]} {input[2]} | \
+#        samtools view -@ {THREADS} -bS - | samtools view -@ {THREADS} -uf 4 - | \
+#        bamToFastq -i stdin -fq {output[2]}
+#        rm -rf $BUFFER* $TMP_FILE
+#        """
+#
+#rule ASSEMBLY_MT_TRINITY_2:
+#    log:
+#        A_LOG
+#    benchmark:
+#        "%s/benchmarks/ASSEMBLY_MT_ASSEMBLY_2.json" % A_OUT
+#    input:
+#        expand('{dir}/{name}', name=[
+#            'MT.R1.unmapped.fq',
+#            'MT.R2.unmapped.fq',
+#            'MT.SE.unmapped.fq'], dir=A_OUT)
+#    output:
+#        '{dir}/MT.assembly_2/final.contigs.fa'.format(dir=A_OUT)
+#    params:
+#        outdir = "{dir}/MT.assembly_2".format(dir=A_OUT)
+#    shell:
+#        """
+#        Trinity --seqType fq\
+#        --left {input[0]} --right {input[1]} --single {input[2]}\
+#        --output {output[0]}\
+#        --JM {MEMTOTAL}G --CPU {THREADS}\
+#        --inchworm_cpu {THREADS} --bflyHeapSpaceMax {MEMTOTAL}G\
+#        --bflyCPU {THREADS} --bflyCalculateCPU\
+#        --normalize_max_read_cov 25  --full_cleanup\
+#        """
+#
+#rule ASSEMBLY_MT_CAT_TRINITY:
+#    log:
+#        A_LOG
+#    benchmark:
+#        "%s/benchmarks/ASSEMBLY_MT_CAT_ASSEMBLY.json" % A_OUT
+#    input:
+#        '{dir}/MT.assembly_1/final.contigs.fa'.format(dir=A_OUT),
+#        '{dir}/MT.assembly_2/final.contigs.fa'.format(dir=A_OUT)
+#    output:
+#        '{dir}/MT.assembly.cat.fa'.format(dir=A_OUT)
+#    shell:
+#        """
+#        cat {input[0]} {input[1]} | awk '/^>/{{print ">contig_MT_" ++i; next}}{{print}}' > {output}
+#        """
+##
 ##################
 # ## MGMT SECTION - ASSEMBLY
 ##################
diff --git a/rules/Assembly/master.rules b/rules/Assembly/master.rules
index 69a28ff..ff8298f 100644
--- a/rules/Assembly/master.rules
+++ b/rules/Assembly/master.rules
@@ -34,9 +34,9 @@ rule ASSEMBLY_MG_ALL:
     shell:
         "echo 'MG Assembly Done'"
 
-#rule ASSEMBLY_FAST:
+#rule ASSEMBLY_MG_FAST:
 #    input:
-#        "%s/MGMT.assembly.merged.fa" % A_OUT
+#        "%s/MG.assembly.merged.fa" % A_OUT
 #    shell:
-#        "echo 'MGMT fast assembly Done'"
+#        "echo 'MG fast assembly done'"
 #
-- 
GitLab