Merge branch 'feature/make_cff' into enhancement/metafusion_refs

mskcc · Aug 4, 2023 · c2c8b70 · c2c8b70
2 parents de72cc2 + d845e07
commit c2c8b70
Show file tree

Hide file tree

Showing 7 changed files with 27 additions and 74 deletions.
diff --git a/bin/Metafusion_forte.sh b/bin/Metafusion_forte.sh
@@ -66,9 +66,13 @@ mkdir $outdir
 #Check CFF file format:
 #Remove entries with nonconformming chromosome name
 #Remove "." from strand field and replace with "NA"
-cat $cff | awk '$1 ~ /[0-9XY]/ && $4 ~ /[0-9XY]/ ' |  awk 'BEGIN{FS=OFS="\t"} $3 !~ /^[-+]$/{$3="NA"} 1' | awk 'BEGIN{FS=OFS="\t"} $6 !~ /^[-+]$/{$6="NA"} 1'   > $outdir/$(basename $cff).reformat
-cff=$outdir/$(basename $cff).reformat
+#cat $cff | awk '$1 ~ /[0-9XY]/ && $4 ~ /[0-9XY]/ ' |  awk 'BEGIN{FS=OFS="\t"} $3 !~ /^[-+]$/{$3="NA"} 1' | awk 'BEGIN{FS=OFS="\t"} $6 !~ /^[-+]$/{$6="NA"} 1'   > $outdir/$(basename $cff).reformat
+#cff=$outdir/$(basename $cff).reformat
 
+all_gene_bed_chrs=`awk -F '\t' '{print $1}' $gene_bed | sort | uniq | sed 's/chr//g '`
+awk -F " " -v arr="${all_gene_bed_chrs[*]}" 'BEGIN{OFS = "\t"; split(arr,arr1); for(i in arr1) dict[arr1[i]]=""} $1 in dict && $4 in dict' $cff >  $outdir/$(basename $cff).cleaned_chr
+grep -v -f $outdir/$(basename $cff).cleaned_chr $cff > problematic_chromosomes.cff
+cff=$outdir/$(basename $cff).cleaned_chr
 
 #Rename cff
 echo Rename cff
@@ -77,9 +81,9 @@ cff=$outdir/$(basename $cff).renamed
 
 ### remove any chromosomes that do not exist in genebed
 all_gene_bed_chrs=`awk -F '\t' '{print $1}' $gene_bed | sort | uniq `
-awk -F " " -v arr="${all_gene_bed_chrs[*]}" 'BEGIN{OFS = "\t"; split(arr,arr1); for(i in arr1) dict[arr1[i]]=""} $1 in dict && $4 in dict' $cff >  $outdir/$(basename $cff).cleaned_chr
-grep -v -f $outdir/$(basename $cff).cleaned_chr $cff > problematic_chromosomes.cff
-cff=$outdir/$(basename $cff).cleaned_chr
+#awk -F " " -v arr="${all_gene_bed_chrs[*]}" 'BEGIN{OFS = "\t"; split(arr,arr1); for(i in arr1) dict[arr1[i]]=""} $1 in dict && $4 in dict' $cff >  $outdir/$(basename $cff).cleaned_chr
+#grep -v -f $outdir/$(basename $cff).cleaned_chr $cff > problematic_chromosomes.cff
+#cff=$outdir/$(basename $cff).cleaned_chr
 
 #Annotate cff
 if [ $genome_fasta ]; then

diff --git a/conf/modules.config b/conf/modules.config
@@ -180,6 +180,10 @@ process {
         ]
     }
 
+    withName: METAFUSION {
+        ext.args = "--numtools ${params.fusion_tool_cutoff}"
+    }
+
     withName: AGFUSION_BATCH {
         ext.args = "-a cff_ext --noncanonical --middlestar"
         publishDir = [

diff --git a/main_test.nf b/main_test.nf
diff --git a/modules/local/convert_to_cff/main.nf b/modules/local/convert_to_cff/main.nf
@@ -2,7 +2,7 @@ process TO_CFF {
     tag "$meta.id"
     label "process_single"
 
-/// must be using singularity 3.7+
+//  must be using singularity 3.7+
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
         'ghcr.io/rocker-org/devcontainer/tidyverse:4' :
         'ghcr.io/rocker-org/devcontainer/tidyverse:4' }"
@@ -11,8 +11,8 @@ process TO_CFF {
     tuple val(meta), val(caller), path(fusions)
 
     output:
-    tuple val(meta), path("*.cff")               , emit: cff
-    path "versions.yml"                          , emit: versions
+    tuple val(meta), path("*.cff"), emit: cff
+    path "versions.yml"           , emit: versions
 
     when:
     task.ext.when == null || task.ext.when

diff --git a/modules/local/metafusion/main.nf b/modules/local/metafusion/main.nf
@@ -12,12 +12,10 @@ process METAFUSION {
     path info
     path fasta
     path blocklist
-    val numtools
 
     output:
     tuple val(meta), path("*final*cluster")             , emit: cluster
     tuple val(meta), path("*.filtered.cff")             , emit: filtered
-    tuple val(meta), path("*reformat*")                 , emit: all
     tuple val(meta), path("cis-sage.cluster")           , emit: cis
     tuple val(meta), path("problematic_chromosomes.cff"), emit: problem_chrom
     path "versions.yml"                                 , emit: versions
@@ -26,9 +24,17 @@ process METAFUSION {
     task.ext.when == null || task.ext.when
 
     script:
+    args = task.ext.args ?: ""
     def sample = "${meta.sample}"
     """
-    Metafusion_forte.sh  --cff $cff --outdir .  --gene_bed $genebed --gene_info $info  --genome_fasta $fasta --recurrent_bedpe $blocklist --num_tools=$numtools
+    Metafusion_forte.sh \\
+        --cff $cff \\
+        --outdir . \\
+        --gene_bed $genebed \\
+        --gene_info $info \\
+        --genome_fasta $fasta \\
+        --recurrent_bedpe $blocklist \\
+        ${args}
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":

diff --git a/nextflow.config b/nextflow.config
@@ -33,7 +33,7 @@ params {
     cosmic_usr                 = null
     fusion_report_cutoff       = 1
     skip_fusion_report         = false
-    numtools                   = 1
+    fusion_tool_cutoff         = 1
 
     // QC
     // rseqc_modules can include ['bam_stat','inner_distance','infer_experiment','junction_annotation','junction_saturation','read_distribution','read_duplication','tin']

diff --git a/subworkflows/local/fusion.nf b/subworkflows/local/fusion.nf
@@ -29,7 +29,6 @@ workflow FUSION {
     gene_bed = params.metafusion_gene_bed
     gene_info = params.metafusion_gene_info
     blocklist = params.metafusion_blocklist
-    numtools = params.numtools
 
     STAR_FOR_ARRIBA(
         reads,
@@ -104,8 +103,7 @@ workflow FUSION {
         gene_bed,
         gene_info,
         fasta,
-        blocklist,
-        numtools
+        blocklist
     )
 
     ADD_FLAG(