1.5.0

.test/config/config.yaml

@@ -3,3 +3,6 @@ samples: config/samples.csv
 
 # Optional path to a `genome.csv` file
 genomes: config/genomes.csv
+
+params:
+  fair_gatk_mutect2_gatk_mutect2_call: "--normal sac_a --enable-all-annotations --max-reads-per-alignment-start 0 --initial-tumor-lod 1.0 --normal-lod 1.0 --tumor-lod-to-emit 1.0 --flow-filter-alleles-qual-threshold 6.0 --flow-filter-alleles-sor-threshold 1.0 --pruning-lod-threshold 0.0009"

.test/config/samples.csv

@@ -1,3 +1,3 @@
-sample_id,upstream_file,downstream_file,species,build,release
-sac_a,data/reads/a.scerevisiae.1.fq,data/reads/a.scerevisiae.2.fq,saccharomyces_cerevisiae,R64-1-1,105
-sac_b,data/reads/b.scerevisiae.1.fq,,saccharomyces_cerevisiae,R64-1-1,105
+sample_id,upstream_file,downstream_file,species,build,release,normal_sample_id
+sac_a,data/reads/a.scerevisiae.1.fq,data/reads/a.scerevisiae.2.fq,saccharomyces_cerevisiae,R64-1-1,105,
+sac_b,data/reads/b.scerevisiae.1.fq,data/reads/b.scerevisiae.2.fq,saccharomyces_cerevisiae,R64-1-1,105,sac_a

.test/makefile

@@ -23,7 +23,7 @@ snakefiles := \
 
 # List of python files to lint
 python_scripts := \
-	"../workflow/scripts/fair_gatk_mutect_germline_multiqc_config.py" 
+	"../workflow/scripts/fair_gatk_mutect2_multiqc_config.py" 
 
 
 
@@ -32,39 +32,33 @@ all: report.txt
 
 
 dag.png: format.txt linter.txt
-	mamba activate test-snakemake-wrappers && \
 	snakemake -s ../workflow/Snakefile \
 		--rulegraph --forceall | dot -Tpng > dag.png && \
 	cp --verbose dag.png ../dag.png
 
 
 report.txt: pipeline.txt
-	mamba activate test-snakemake-wrappers && \
 	snakemake -s ../workflow/Snakefile --profile local \
 		--report report.zip 2>&1 | tee report.txt
 
 
 pipeline.txt: linter.txt
-	mamba activate test-snakemake-wrappers && \
 	snakemake -s ../workflow/Snakefile \
 		--cores 7 \
 		--software-deployment-method conda \
 		2>&1 | tee pipeline.txt
 
 
 format.txt:
-	mamba activate test-snakemake-wrappers && \
 	snakefmt $(snakefiles) 2>&1 | tee format.txt
 
 
 linter.txt: format.txt black.txt
-	mamba activate test-snakemake-wrappers && \
 	snakemake -s ../workflow/Snakefile \
 		--lint 2>&1 | tee linter.txt
 
 
 black.txt:
-	mamba activate test-snakemake-wrappers && \
 	black $(python_scripts) 2>&1 | tee black.txt
 
 

CHANGELOG.md

@@ -1,3 +1,9 @@
+# 1.4.2
+
+## Features:
+
+* More time for Mutect2 rule
+
 # 1.4.1
 
 ## Features:

README.md

@@ -1,22 +1,22 @@
 [![Snakemake](https://img.shields.io/badge/snakemake-≥7.29.0-brightgreen.svg)](https://snakemake.github.io)
-[![GitHub actions status](https://github.com/tdayris/fair_gatk_mutect_germline/workflows/Tests/badge.svg)](https://github.com/tdayris/fair_gatk_mutect_germline/actions?query=branch%3Amain+workflow%3ATests)
+[![GitHub actions status](https://github.com/tdayris/fair_gatk_mutext2/workflows/Tests/badge.svg)](https://github.com/tdayris/fair_gatk_mutext2/actions?query=branch%3Amain+workflow%3ATests)
 
 Do not use. Active dev.
 
 Snakemake workflow used to call germline variants with GATK-Mutect2
 
 ## Usage
 
-The usage of this workflow is described in the [Snakemake workflow catalog](https://snakemake.github.io/snakemake-workflow-catalog?usage=tdayris/fair_gatk_mutect_germline) 
-it is also available [locally](https://github.com/tdayris/fair_gatk_mutect_germline/blob/main/workflow/report/usage.rst) on a single page.
+The usage of this workflow is described in the [Snakemake workflow catalog](https://snakemake.github.io/snakemake-workflow-catalog?usage=tdayris/fair_gatk_mutext2) 
+it is also available [locally](https://github.com/tdayris/fair_gatk_mutext2/blob/main/workflow/report/usage.rst) on a single page.
 
 ## Results
 
-A complete description of the results can be found here in [workflow reports](https://github.com/tdayris/fair_gatk_mutect_germline/blob/main/workflow/report/results.rst).
+A complete description of the results can be found here in [workflow reports](https://github.com/tdayris/fair_gatk_mutext2/blob/main/workflow/report/results.rst).
 
 ## Material and Methods
 
-The tools used in this pipeline are described [here](https://github.com/tdayris/fair_gatk_mutect_germline/blob/main/workflow/report/material_methods.rst) textually. Web-links are available below:
+The tools used in this pipeline are described [here](https://github.com/tdayris/fair_gatk_mutext2/blob/main/workflow/report/material_methods.rst) textually. Web-links are available below:
 
 ![workflow_rulegraph](dag.png)
 

config/samples.csv

@@ -1,2 +1,4 @@
-sample_id,upstream_file,downstream_file,species,build,release
-sac_a,ngs-test-data/reads/a.scerevisiae.1.fq,ngs-test-data/reads/a.scerevisiae.2.fq,saccharomyces_cerevisiae,R64-1-1,105
+sample_id,upstream_file,downstream_file,species,build,release,normal_sample_id
+sac_a,ngs-test-data/reads/a.scerevisiae.1.fq,ngs-test-data/reads/a.scerevisiae.2.fq,saccharomyces_cerevisiae,R64-1-1,105,
+sac_b,ngs-test-data/reads/b.scerevisiae.1.fq,,saccharomyces_cerevisiae,R64-1-1,105,
+sac_c,ngs-test-data/reads/b.scerevisiae.1.fq,ngs-test-data/reads/b.scerevisiae.2.fq,saccharomyces_cerevisiae,R64-1-1,105,sac_a

workflow/rules/bcftools.smk

@@ -1,61 +1,89 @@
+rule fair_gatk_mutect2_bcftools_norm_split_multiallelic:
+    input:
+        "tmp/fair_gatk_mutect2_snpsift_vartype/{species}.{build}.{release}.{datatype}/{sample}.vcf",
+        ref=lambda wildcards: select_fasta(wildcards),
+    output:
+        temp(
+            "tmp/fair_gatk_mutect2_bcftools_norm_split_multiallelic/{species}.{build}.{release}.{datatype}/{sample}.vcf.gz",
+        ),
+    threads: 1
+    resources:
+        mem_mb=lambda wildcards, attempt: attempt * 3_000,
+        runtime=lambda wildcards, attempt: attempt * 25,
+        tmpdir=tmp,
+    log:
+        "logs/fair_gatk_mutect2_bcftools_norm_split_multiallelic/{species}.{build}.{release}.{datatype}/{sample}.log",
+    benchmark:
+        "benchmark/fair_gatk_mutect2_bcftools_norm_split_multiallelic/{species}.{build}.{release}.{datatype}/{sample}.tsv"
+    params:
+        extra=lookup_config(
+            dpath="params/fair_gatk_mutect2_bcftools_norm_split_multiallelic",
+            default="--multiallelics -any --rm-dup none",
+        ),
+    wrapper:
+        f"{snakemake_wrappers_prefix}/bio/bcftools/norm"
+
+
 rule fair_gatk_mutect2_bcftools_view:
     input:
-        "tmp/fair_gatk_mutect2/snpsift_vartype/{species}.{build}.{release}.{datatype}/{sample}.vcf",
+        "tmp/fair_gatk_mutect2_bcftools_norm_split_multiallelic/{species}.{build}.{release}.{datatype}/{sample}.vcf.gz",
     output:
         protected(
-            "results/{species}.{build}.{release}.{datatype}/VariantCalling/Germline/VCF/{sample}.vcf.gz"
+            "results/{species}.{build}.{release}.{datatype}/VariantCalling/VCF/{sample}.vcf.gz"
         ),
     threads: 1
     resources:
-        mem_mb=lambda wildcards, attempt: attempt * 1024 * 3,
+        mem_mb=lambda wildcards, attempt: attempt * 3_000,
         runtime=lambda wildcards, attempt: attempt * 15,
         tmpdir=tmp,
     log:
-        "logs/fair_gatk_mutect2/bcftools_view/{species}.{build}.{release}.{datatype}/{sample}.log",
+        "logs/fair_gatk_mutect2i_bcftools_view/{species}.{build}.{release}.{datatype}/{sample}.log",
     benchmark:
-        "benchmark/fair_gatk_mutect2/bcftools_view/{species}.{build}.{release}.{datatype}/{sample}.tsv"
+        "benchmark/fair_gatk_mutect2_bcftools_view/{species}.{build}.{release}.{datatype}/{sample}.tsv"
     params:
-        extra=lookup_config(dpath="params/fair_gatk_mutect2/bcftools/view", default=""),
+        extra=lookup_config(
+            dpath="params/fair_gatk_mutect2_bcftools_view", default="--with-header"
+        ),
     wrapper:
         f"{snakemake_wrappers_prefix}/bio/bcftools/view"
 
 
 rule fair_gatk_mutect2_bcftools_index:
     input:
-        "results/{species}.{build}.{release}.{datatype}/VariantCalling/Germline/VCF/{sample}.vcf.gz",
+        "results/{species}.{build}.{release}.{datatype}/VariantCalling/VCF/{sample}.vcf.gz",
     output:
-        "results/{species}.{build}.{release}.{datatype}/VariantCalling/Germline/VCF/{sample}.vcf.gz.tbi",
+        "results/{species}.{build}.{release}.{datatype}/VariantCalling/VCF/{sample}.vcf.gz.tbi",
     threads: 1
     resources:
-        mem_mb=lambda wildcards, attempt: attempt * 1024 * 2,
+        mem_mb=lambda wildcards, attempt: attempt * 2_000,
         runtime=lambda wildcards, attempt: attempt * 15,
         tmpdir=tmp,
     log:
-        "logs/fair_gatk_mutect2/bcftools_index/{species}.{build}.{release}.{datatype}/{sample}.log",
+        "logs/fair_gatk_mutect2_bcftools_index/{species}.{build}.{release}.{datatype}/{sample}.log",
     benchmark:
-        "benchmark/fair_gatk_mutect2/bcftools_index/{species}.{build}.{release}.{datatype}/{sample}.tsv"
+        "benchmark/fair_gatk_mutect2_bcftools_index/{species}.{build}.{release}.{datatype}/{sample}.tsv"
     wrapper:
         f"{snakemake_wrappers_prefix}/bio/bcftools/index"
 
 
 rule fair_gatk_mutect2_bcftools_mutect2_stats:
     input:
-        "results/{species}.{build}.{release}.{datatype}/VariantCalling/Germline/VCF/{sample}.vcf.gz",
-        "results/{species}.{build}.{release}.{datatype}/VariantCalling/Germline/VCF/{sample}.vcf.gz.tbi",
+        "results/{species}.{build}.{release}.{datatype}/VariantCalling/VCF/{sample}.vcf.gz",
+        "results/{species}.{build}.{release}.{datatype}/VariantCalling/VCF/{sample}.vcf.gz.tbi",
     output:
         temp(
-            "tmp/fair_gatk_mutect2/bcftools_mutect2_stats/{species}.{build}.{release}.{datatype}/{sample}.stats.txt"
+            "tmp/fair_gatk_mutect2_bcftools_mutect2_stats/{species}.{build}.{release}.{datatype}/{sample}.stats.txt"
         ),
     threads: 2
     resources:
         mem_mb=lambda wildcards, attempt: (1024 * 7) * attempt,
         runtime=lambda wildcards, attempt: 30 * attempt,
         tmpdir=tmp,
     log:
-        "logs/fair_gatk_mutect2/bcftools_mutect2_stats/{species}.{build}.{release}.{datatype}/{sample}.log",
+        "logs/fair_gatk_mutect2_bcftools_mutect2_stats/{species}.{build}.{release}.{datatype}/{sample}.log",
     benchmark:
-        "benchmark/fair_gatk_mutect2/bcftools_mutect2_stats/{species}.{build}.{release}.{datatype}/{sample}.tsv"
+        "benchmark/fair_gatk_mutect2_bcftools_mutect2_stats/{species}.{build}.{release}.{datatype}/{sample}.tsv"
     params:
-        lookup_config(dpath="params/fair_gatk_mutect2/bcftools/stats", default=""),
+        lookup_config(dpath="params/fair_gatk_mutect2_bcftools_stats", default=""),
     wrapper:
         f"{snakemake_wrappers_prefix}/bio/bcftools/stats"