Merge pull request #4 from davidebolo1993/impg_initial

merging impg_initial into master

.gitignore

@@ -9,4 +9,5 @@ cosigt_smk/.snakemake/*
 cosigt_smk/workflow/scripts/__pycache__
 cosigt_smk/snakemake.run.sh
 cosigt_smk/config/*
+cosigt_smk/benchmarks/*
 cosigt_smk/.*

cosigt_smk/README.md

@@ -1,3 +1 @@
-# Snakemake pipeline
-
-Docs are in preparation
+# Snakemake pipeline

cosigt_smk/workflow/Snakefile

@@ -1,4 +1,5 @@
 import pandas as pd
+from glob import glob
 
 configfile: 'config/config.yaml'
 
@@ -7,7 +8,10 @@ df=(pd.read_table(config['samples'], dtype={'sample_id': str, 'cram':str})
 	.sort_index()
 )
 
-include: 'rules/extract.smk'
+include: 'rules/impg.smk'
+include: 'rules/bedtools.smk'
+include: 'rules/samtools.smk'
+include: 'rules/faidx.smk'
 include: 'rules/pggb.smk'
 include: 'rules/odgi.smk'
 include: 'rules/bwa-mem2.smk'

cosigt_smk/workflow/rules/bedtools.smk

@@ -0,0 +1,26 @@
+rule bedtools_merge:
+	'''
+	https://github.com/arq5x/bedtools2
+	'''
+	input:
+		rules.impg_project.output
+	output:
+		config['output'] + '/bedtools/{region}.bed'
+	threads:
+		1
+	resources:
+		mem_mb=lambda wildcards, attempt: attempt * config['default']['mem_mb'],
+		time=lambda wildcards, attempt: attempt * config['default']['time']
+	container:
+		'docker://davidebolo1993/cosigt_workflow:latest'
+	benchmark:
+		'benchmarks/{region}.bedtools_merge.benchmark.txt'
+
+	shell:
+		'''
+		bedtools sort \
+		-i {input} | \
+		bedtools merge \
+		-d 50000 \
+		-i - > {output}
+		'''	

cosigt_smk/workflow/rules/bwa-mem2.smk

@@ -1,30 +1,33 @@
-rule bwa_mem2_index:
+rule bwamem2_index:
 	'''
-	bwa index
+	https://github.com/bwa-mem2/bwa-mem2
 	'''
 	input:
-		rules.odgi_paths_fasta.output
+		rules.pyfaidx_extract.output
 	output:
-		multiext(config['output'] + '/odgi/paths/fasta/{region}.fa', '.bwt.2bit.64', '.pac', '.ann', '.amb', '.0123')	
+		multiext(config['output'] + '/pyfaidx/{region}.fasta', '.bwt.2bit.64', '.pac', '.ann', '.amb', '.0123')	
 	threads:
 		1
 	resources:
 		mem_mb=lambda wildcards, attempt: attempt * config['bwa-mem2']['mem_mb'],
 		time=lambda wildcards, attempt: attempt * config['bwa-mem2']['time']
 	container:
 		'docker://davidebolo1993/cosigt_workflow:latest'
+	benchmark:
+		'benchmarks/{region}.bwamem2_index.benchmark.txt'
 	shell:
 		'''
 		bwa-mem2 index {input}
 		'''
 
-rule bwa_mem2_samtools_sort:
+rule bwamem2_mem_samtools_sort:
 	'''
-	bwa-mem2 and sam-to-bam conversion with samtools
+	https://github.com/bwa-mem2/bwa-mem2
+	https://github.com/samtools/samtools
 	'''
 	input:
-		ref=rules.odgi_paths_fasta.output,
-		idx=rules.bwa_mem2_index.output,
+		ref=rules.pyfaidx_extract.output,
+		idx=rules.bwamem2_index.output,
 		sample=rules.samtools_fasta.output
 	output:
 		config['output'] + '/bwa-mem2/{sample}/{region}.realigned.bam'
@@ -35,7 +38,14 @@ rule bwa_mem2_samtools_sort:
 		time=lambda wildcards, attempt: attempt * config['bwa-mem2']['time']
 	container:
 		'docker://davidebolo1993/cosigt_workflow:latest'
+	benchmark:
+		'benchmarks/{sample}.{region}.bwamem2_mem_samtools_sort.benchmark.txt'
 	shell:
 		'''
-		bwa-mem2 mem -t {threads} {input.ref} {input.sample} | samtools sort -@ {threads} - > {output}
+		bwa-mem2 mem \
+		-t {threads} \
+		{input.ref} \
+		{input.sample} | samtools sort \
+		-@ {threads} \
+		- > {output}
 		'''

cosigt_smk/workflow/rules/cosigt.smk

@@ -1,11 +1,11 @@
 rule cosigt_genotype:
 	'''
-	cosigt genotype
+	https://github.com/davidebolo1993/cosigt
 	'''
 	input:
 		zpath=rules.odgi_paths_matrix.output,
 		xpack=rules.gafpack_coverage.output,
-		cluster=rules.cluster.output
+		cluster=rules.make_clusters.output
 	output:
 		geno=config['output'] + '/cosigt/{sample}/{region}/cosigt_genotype.tsv',
 		combo=config['output'] + '/cosigt/{sample}/{region}/sorted_combos.tsv'
@@ -18,7 +18,14 @@ rule cosigt_genotype:
 		'docker://davidebolo1993/cosigt_workflow:latest'
 	params:
 		prefix=config['output'] + '/cosigt/{sample}/{region}'
+	benchmark:
+		'benchmarks/{sample}.{region}.cosigt_genotype.benchmark.txt'
 	shell:
 		'''
-		cosigt -p {input.zpath} -g {input.xpack} -c {input.cluster} -b resources/extra/blacklist.txt -o {params.prefix}
+		cosigt \
+		-p {input.zpath} \
+		-g {input.xpack} \
+		-c {input.cluster} \
+		-b resources/extra/blacklist.txt \
+		-o {params.prefix}
 		'''

cosigt_smk/workflow/rules/faidx.smk

@@ -0,0 +1,46 @@
+rule pyfaidx_extract:
+	'''
+	https://github.com/mdshw5/pyfaidx/
+	'''
+	input:
+		fasta=config['assemblies'],
+		bed=rules.bedtools_merge.output
+	output:
+		config['output'] + '/pyfaidx/{region}.fasta'
+	threads:
+		1
+	resources:
+		mem_mb=lambda wildcards, attempt: attempt * config['default']['mem_mb'],
+		time=lambda wildcards, attempt: attempt * config['default']['time']
+	container:
+		'docker://davidebolo1993/cosigt_workflow:latest'
+	benchmark:
+		'benchmarks/{region}.pyfaidx_extract.benchmark.txt'
+	shell:
+		'''
+		faidx \
+		--bed {input.bed} \
+		{input.fasta} > {output}
+		'''
+
+rule samtools_faidx_index:
+	'''
+	https://github.com/samtools/samtools
+	'''
+	input:
+		rules.pyfaidx_extract.output
+	output:
+		config['output'] + '/pyfaidx/{region}.fasta.fai'
+	threads:
+		1
+	resources:
+		mem_mb=lambda wildcards, attempt: attempt * config['default']['mem_mb'],
+		time=lambda wildcards, attempt: attempt * config['default']['time']
+	container:
+		'docker://davidebolo1993/cosigt_workflow:latest'
+	benchmark:
+		'benchmarks/{region}.samtools_faidx_index.benchmark.txt'
+	shell:
+		'''
+		samtools faidx {input}
+		'''