From f2dfbb2b312d56901acb95f4dbba5584a4311096 Mon Sep 17 00:00:00 2001
From: Helene Rimbert <helene.rimbert@inra.fr>
Date: Wed, 18 Nov 2020 15:09:51 +0100
Subject: [PATCH] reorder query annotation related rules from main Snakefile to
rules/preprocessGenomes.smk
---
Snakefile | 30 +++++++-----------------------
rules/preprocessGenomes.smk | 24 +++++++++++++++++++++++-
2 files changed, 30 insertions(+), 24 deletions(-)
diff --git a/Snakefile b/Snakefile
index 9d727b7..5e0fc99 100644
--- a/Snakefile
+++ b/Snakefile
@@ -15,35 +15,19 @@ rule all:
config['finalPrefix']+'_LC.proteins.fasta',
config['finalPrefix']+'_gmap_differentChrom.gff3',
config['finalPrefix']+'_anchoringSummary.csv',
- config['finalPrefix']+'_blatSummary.csv'
+ config['finalPrefix']+'_blatSummary.csv',
+ 'report/dag.png',
+ 'report/rulegraph.png'
-rule grepGffFeature:
- message: " Collect selected features from GFF file"
- input: config['annotationQuery']
- params: config['featureType']
- output: temp(config['results']+"/1.features.bed")
- log: config['results']+"/1.grepGffFeature.log"
- shell:
- """
- bin/gff2bed.sh {params} {input} 1> {output} 2> {log}
- """
-rule splitGffPerChrom:
- message: "Split Gff Features per chromosome: current is {wildcards.chrom}"
- input: config['results']+"/1.features.bed"
- output: temp(config['results']+"/1.features/{chrom}.bed")
- log: config['results']+"/1.features/{chrom}.fgrep.log"
- params: "chr{chrom}"
- shell:
- """
- fgrep {params} {input} 1> {output} 2> {log}
- """
rule createDiagrams:
message: "Create dag and rulegraph of the pipeline"
+ output: dag='report/dag.png', rulegraphe='report/rulegraph.png'
shell:
"""
- snakemake --dag |dot -T png > report/dag.png
- snakemake --rulegraph |dot -T png > report/rulegraph.png
+ snakemake --dag |dot -T png > {output.dag}
+ snakemake --rulegraph |dot -T png > {output.rulegraph}
"""
+
include: "rules/preprocessISBP.smk"
include: "rules/preprocessGenomes.smk"
include: "rules/bedtoolsClosest.smk"
diff --git a/rules/preprocessGenomes.smk b/rules/preprocessGenomes.smk
index b138847..5c67eec 100644
--- a/rules/preprocessGenomes.smk
+++ b/rules/preprocessGenomes.smk
@@ -1,3 +1,25 @@
+rule grepGffFeature:
+ message: " Collect selected features from GFF file"
+ input: config['annotationQuery']
+ params: config['featureType']
+ output: temp(config['results']+"/1.features.bed")
+ log: config['results']+"/1.grepGffFeature.log"
+ shell:
+ """
+ bin/gff2bed.sh {params} {input} 1> {output} 2> {log}
+ """
+
+rule splitGffPerChrom:
+ message: "Split Gff Features per chromosome: current is {wildcards.chrom}"
+ input: config['results']+"/1.features.bed"
+ output: temp(config['results']+"/1.features/{chrom}.bed")
+ log: config['results']+"/1.features/{chrom}.fgrep.log"
+ params: "chr{chrom}"
+ shell:
+ """
+ fgrep {params} {input} 1> {output} 2> {log}
+ """
+
rule indexQuery:
message: " Indexing Query fasta file using samtools faidx"
input: config['queryFasta']
@@ -16,4 +38,4 @@ rule gmapIndexTarget:
output: directory(config['results']+"/target_gmapindex")
params: indexname="target_gmapindex", indexPath=config['results']
log: config['results']+"/target_gmapindex.log"
- shell: "gmap_build -D {params.indexPath} -d {params.indexname} {input} &> {log}"
\ No newline at end of file
+ shell: "gmap_build -D {params.indexPath} -d {params.indexname} {input} &> {log}"
--
GitLab