diff --git a/init.rule b/init.rule
index 00b160665cb2bf05822b259df84ba5d6306d36ab..55944d3a28bd571096c3d7dcc19ea274a8fb698b 100644
--- a/init.rule
+++ b/init.rule
@@ -1,10 +1,13 @@
include:
"config"
-
rule ALL:
input:
- expand("{path}/{filter}.{ext}", path=DBPATH + "/human", filter=config["human_filtering"]["filter"], ext=['fa']),
+ expand(
+ "{path}/{filter}.{ext}", path=DBPATH + "/human",
+ filter=config["human_filtering"]["filter"],
+ ext=['fa', 'fa.amb', 'fa.ann', 'fa.bwt', 'fa.pac', 'fa.sa']
+ ),
expand("{path}/{files}.fasta", files=config["sortmerna"]["files"], path=DBPATH + "/sortmerna"),
expand("{path}/{db}", path=DBPATH, db=config["prokka"]["databases"]),
"%s/adapters/adapters.done" % DBPATH,
diff --git a/rules/Assembly/MGMT.rules b/rules/Assembly/MGMT.rules
index 33f766de00189d14016599bd115a223cde279280..c7a79f12b4512cc1102df2b084d94c789d992049 100644
--- a/rules/Assembly/MGMT.rules
+++ b/rules/Assembly/MGMT.rules
@@ -3,7 +3,7 @@
###################################################################################################
#########################
-## ## MT MEGAHIT ASSEMBLY
+# ## MT MEGAHIT ASSEMBLY
#########################
rule ASSEMBLY_MT_MEGAHIT_1:
@@ -110,108 +110,6 @@ rule ASSEMBLY_MT_CAT_MEGAHIT:
"""
cat {input[0]} {input[1]} | awk '/^>/{{print ">contig_MT_" ++i; next}}{{print}}' > {output}
"""
-#########################
-## ## MT TRINITY ASSEMBLY
-#########################
-#
-#rule ASSEMBLY_MT_TRINITY_TRINITY_1:
-# log:
-# A_LOG
-# benchmark:
-# "%s/benchmarks/ASSEMBLY_MT_TRINITY_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_TRINITY_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_TRINITY_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_TRINITY.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 ff8298fbe38e216702eaef92cf1470d2f0f1730c..75eebedbdb9a641a6266aad23a58295eba8c6408 100644
--- a/rules/Assembly/master.rules
+++ b/rules/Assembly/master.rules
@@ -7,6 +7,7 @@ include:
include:
"MG.rules"
+
def assembly_output_files():
return expand('{dir}/{name}', name=[
'MG.reads.sorted.bam',
@@ -33,10 +34,3 @@ rule ASSEMBLY_MG_ALL:
"%s/MG.assembly.merged.fa" % A_OUT
shell:
"echo 'MG Assembly Done'"
-
-#rule ASSEMBLY_MG_FAST:
-# input:
-# "%s/MG.assembly.merged.fa" % A_OUT
-# shell:
-# "echo 'MG fast assembly done'"
-#
diff --git a/rules/Preprocessing/MG.rules b/rules/Preprocessing/MG.rules
index 62dcaebb1b22963f3c74c91f1c38d5a796b43b57..0b230b13ac46d30edda9fb75bb5ebb552da33124 100644
--- a/rules/Preprocessing/MG.rules
+++ b/rules/Preprocessing/MG.rules
@@ -39,7 +39,15 @@ rule PREPROCESSING_MG_TRIM:
'MG.SE2.uniq.trimmed.fq'], dir=P_OUT)
shell:
"""
- java -jar {config[trimmomatic][jarfile]} PE -threads {THREADS} {input[0]} {input[1]} {output} ILLUMINACLIP:{DBPATH}/adapters/{config[trimmomatic][adapter]}-PE.fa:{config[trimmomatic][seed_mismatch]}:{config[trimmomatic][palindrome_clip_threshold]}:{config[trimmomatic][simple_clip_threshold]} LEADING:{config[trimmomatic][leading]} TRAILING:{config[trimmomatic][trailing]} SLIDINGWINDOW:{config[trimmomatic][window_size]}:{config[trimmomatic][window_quality]} MINLEN:{config[trimmomatic][minlen]} MAXINFO:{config[trimmomatic][target_length]}:{config[trimmomatic][strictness]} >> {log} 2>&1
+ java -jar {config[trimmomatic][jarfile]} PE -threads {THREADS} {input[0]} {input[1]} {output} \
+ ILLUMINACLIP:{DBPATH}/adapters/{config[trimmomatic][adapter]}-PE.fa:\
+ {config[trimmomatic][seed_mismatch]}:{config[trimmomatic][palindrome_clip_threshold]}:\
+ {config[trimmomatic][simple_clip_threshold]} \
+ LEADING:{config[trimmomatic][leading]} \
+ TRAILING:{config[trimmomatic][trailing]} \
+ SLIDINGWINDOW:{config[trimmomatic][window_size]}:{config[trimmomatic][window_quality]} \
+ MINLEN:{config[trimmomatic][minlen]} \
+ MAXINFO:{config[trimmomatic][target_length]}:{config[trimmomatic][strictness]} >> {log} 2>&1
"""
benchmark:
"%s/benchmarks/PREPROCESSING_MG_TRIM.json" % P_OUT
diff --git a/rules/Preprocessing/MT.rules b/rules/Preprocessing/MT.rules
index d305bfcd81a84b9f314f46789ebeeeaedeec1799..990a1596e6fcb2603faee373f974a96da2e8a070 100644
--- a/rules/Preprocessing/MT.rules
+++ b/rules/Preprocessing/MT.rules
@@ -22,7 +22,13 @@ rule PREPROCESSING_MT_TRIM:
'MT.SE2.trimmed.fq'], dir=P_OUT)
shell:
"""
- java -jar {config[trimmomatic][jarfile]} PE -threads {THREADS} {input[0]} {input[1]} {output} ILLUMINACLIP:{DBPATH}/adapters/{config[trimmomatic][adapter]}-PE.fa:{config[trimmomatic][seed_mismatch]}:{config[trimmomatic][palindrome_clip_threshold]}:{config[trimmomatic][simple_clip_threshold]} LEADING:{config[trimmomatic][leading]} TRAILING:{config[trimmomatic][trailing]} SLIDINGWINDOW:{config[trimmomatic][window_size]}:{config[trimmomatic][window_quality]} MINLEN:{config[trimmomatic][minlen]} MAXINFO:{config[trimmomatic][target_length]}:{config[trimmomatic][strictness]} >> {log} 2>&1
+ java -jar {config[trimmomatic][jarfile]} PE -threads {THREADS} {input[0]} {input[1]} {output} \
+ ILLUMINACLIP:{DBPATH}/adapters/{config[trimmomatic][adapter]}-PE.fa:{config[trimmomatic][seed_mismatch]}:\
+ {config[trimmomatic][palindrome_clip_threshold]}:{config[trimmomatic][simple_clip_threshold]} \
+ LEADING:{config[trimmomatic][leading]} TRAILING:{config[trimmomatic][trailing]} \
+ SLIDINGWINDOW:{config[trimmomatic][window_size]}:{config[trimmomatic][window_quality]} \
+ MINLEN:{config[trimmomatic][minlen]} \
+ MAXINFO:{config[trimmomatic][target_length]}:{config[trimmomatic][strictness]} >> {log} 2>&1
"""
benchmark:
"%s/benchmarks/PREPROCESSING_MT_TRIM.json" % P_OUT
@@ -44,7 +50,6 @@ rule PREPROCESSING_MT_CAT_TRIMMED_SE:
"cat {input[0]} {input[1]} > {output}"
-
rule PREPROCESSING_MT_FILTER_RRNA:
log:
P_LOG
@@ -81,10 +86,12 @@ rule PREPROCESSING_MT_FILTER_RRNA:
shell("""
TMP_R12=$(mktemp --tmpdir={TMPDIR} -t "tmp_XXXXXX.R12")
{config[sortmerna][scripts_path]}/merge-paired-reads.sh {input[0]} {input[1]} $TMP_R12.fq
- sortmerna --ref {ref} --reads $TMP_R12.fq --fastx --aligned $TMP_R12.rrna --other $TMP_R12.rrna_filt -a {THREADS} --paired_in -m {mem}
+ sortmerna --ref {ref} --reads $TMP_R12.fq --fastx --aligned $TMP_R12.rrna \
+ --other $TMP_R12.rrna_filt -a {THREADS} --paired_in -m {mem}
{config[sortmerna][scripts_path]}/unmerge-paired-reads.sh $TMP_R12.rrna.fq {output[3]} {output[4]}
{config[sortmerna][scripts_path]}/unmerge-paired-reads.sh $TMP_R12.rrna_filt.fq {output[0]} {output[1]}
- sortmerna --ref {ref} --reads {input[2]} --fastx --aligned {P_OUT}/MT.SE.trimmed.rna --other {P_OUT}/MT.SE.trimmed.rna_filtered -a {THREADS} -m {mem}
+ sortmerna --ref {ref} --reads {input[2]} --fastx --aligned {P_OUT}/MT.SE.trimmed.rna \
+ --other {P_OUT}/MT.SE.trimmed.rna_filtered -a {THREADS} -m {mem}
rm $TMP_R12*
""")
@@ -119,8 +126,9 @@ rule PREPROCESSING_MT_FILTER_HG:
<(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]}
-
+ samtools view -@ {THREADS} -bF 0x800 - | samtools sort -o -@ {THREADS} -m {MEMCORE}G -n - $BUFFER \
+ | bamToFastq -i stdin -fq {output[0]} -fq2 {output[1]}
+
if [[ -s {input[2]} ]]
then
bwa mem -v 1 -t {THREADS} {input[3]} {input[2]} | samtools view -@ {THREADS} -bS - | \
@@ -129,6 +137,6 @@ rule PREPROCESSING_MT_FILTER_HG:
echo "{input[2]} is empty, skipping single end human sequence filtering, but creating it anyway..."
touch {output[2]}
fi
-
+
rm -rf $BUFFER* $TMP_FILE
"""
diff --git a/rules/Preprocessing/master.rules b/rules/Preprocessing/master.rules
index eb8a26ade9a76d1839fd9633b77e04c473f2c547..36ca9cdf046ba4aec0ef2d3e4a7b58513c9414e2 100644
--- a/rules/Preprocessing/master.rules
+++ b/rules/Preprocessing/master.rules
@@ -15,7 +15,14 @@ def preprocessing_output_files():
"""
Dynamically generate output files names based on parameters
"""
- return [preprocessed_mt('R1'), preprocessed_mt('R2'), preprocessed_mt('SE')] + [preprocessed_mg('R1'), preprocessed_mg('R2'), preprocessed_mg('SE')]
+ return [
+ preprocessed_mt('R1'),
+ preprocessed_mt('R2'),
+ preprocessed_mt('SE'),
+ preprocessed_mg('R1'),
+ preprocessed_mg('R2'),
+ preprocessed_mg('SE')
+ ]
def preprocessed_mg(target, dir=P_OUT):