Merge pull request #11 from NorwegianVeterinaryInstitute/dev_filter_contig

Dev filter contig - fixing small bugs in main

Author: evezeyl

SAGA/run_filter_contigs.sh

@@ -0,0 +1,13 @@
+#!/bin/bash
+# Run filter_contigs track on SAGA
+# Usage: bash SAGA/run_filter_contigs.sh
+
+nextflow run main.nf \
+    -c SAGA/saga_sapro.config \
+    -profile apptainer \
+    --track filter_contigs \
+    --input samples.csv \
+    --out_dir results \
+    --blastDB /cluster/shared/databases/blast/nt \
+    --ranked_taxo_file /path/to/ranked_taxo.tsv \
+    --positive_filter "Oomycetes"

conf/modules_varwrrum.config

@@ -153,7 +153,7 @@ process {
 
 
   // PREP SAMPLES POP   
-    withName: '  VARWRRUM:VAR_PREP_SAMPLES:SAMTOOLS_MERGE_BAMS|VARWRRUM:VAR_PREP_SAMPLES:SAMTOOLS_COORDSORT_INDEX' {
+    withName: 'VARWRRUM:VAR_PREP_SAMPLES:SAMTOOLS_MERGE_BAMS|VARWRRUM:VAR_PREP_SAMPLES:SAMTOOLS_COORDSORT_INDEX' {
         publishDir = [
             [
                 path:   { "${params.out_dir}/versions" },

modules/BWAMEM.nf

@@ -53,26 +53,3 @@ process BWAMEM_VAR {
 // [W::sam_hdr_link_pg] PG line with ID:samtools-5D89A8C2 has a PP link to missing program 'bwa'
 
 
-// Need to find the way to write better 
-process BWAMEM_VAR {
-	conda (params.enable_conda ? 'bioconda::bwa=0.7.8' : null)
-	container 'quay.io/biocontainers/bwa:0.7.8--he4a0461_9'
-
-    label 'process_high'
-    
-    tag "$ID" 
-
-    input:
-    tuple val(ID), path(R1), path(R2), path(assembly)
-
-    output:
-    tuple val(ID), path("*_map1.sam"), emit: shortbam_ch // will need to remove this one - in compass
-    tuple val(ID), path("*_map1.sam"), emit: sam // check actually is not supposed to be compressed
-
-    script:
-    threads = task.cpus * 2
-    """
-	bwa index $assembly 
-	bwa mem -t $threads -v 0 -R '@RG\tID:${ID}\tSM:${ID}' $params.bwamem_args  $assembly $R1 $R2 > ${ID}_map1.sam   
-    """
-}

nextflow.config

@@ -11,7 +11,8 @@ params {
     // ID(name),assembly (path),longreads(path),R1(path),R2(path),type(assembly),comments(for you in case) 
     // options type: short, long, hybrid
 
-    input              = null 
+    input              = null
+    out_dir            = "results"
 
 
     // TRACK Choice : purely_raw (not done), tryssembly (not done), filter_contigs, coverage, haplopurge, varwrrum
@@ -160,9 +161,10 @@ params {
 // Profiles 
 includeConfig 'conf/profiles.config' 
 
-// Modules 
+// Modules
 includeConfig 'conf/base.config'
 // I had to split the modules config - was unpractical
+includeConfig 'conf/modules_filter_contigs.config'
 includeConfig 'conf/modules_compass.config'
 includeConfig 'conf/modules_varwrrum.config'
 

workflows/FILTER_CONTIGS.nf

@@ -6,7 +6,7 @@ workflow FILTER_CONTIGS {
     input_ch = Channel
         .fromPath(params.input, checkIfExists: true)
         .splitCsv(header:['ID', 'assembly'], skip: 1, sep:",", strip:true)
-        .map { row -> (ID, assembly) =  [row.ID, row.assembly]}
+        .map { row -> tuple(row.ID, file(row.assembly)) }
 
     FILTER_CONTIGS_SUB(input_ch)