Simplified consensus workflow

Removed racon, now use only medaka against the input reference

envs/racon-medaka.yaml → envs/medaka.yaml

@@ -1,8 +1,7 @@
-name: racon-medaka
+name: medaka
 channels:
  - conda-forge
  - bioconda
 dependencies:
- - racon
  - medaka
 

main.nf

@@ -129,28 +129,25 @@ workflow {
  CollateCounts(count_ch.counts.collect())
 
  if (params.consensus) {
- // reformat channel for consensus input
- // to tuple (sampleName, bamFile, fastqFile)
- // filter out unmapped and out files from splitcode
  if (params.count_only) {
- count_ch.alignments.filter { sampleName, bamFile, fastqFile ->
- !bamFile.getName().startsWith("unmapped.bam") &&
- !bamFile.getName().startsWith("out.bam")
- }.set{ bam_ch }
+ demux_ch.set{ fastq_ch }
  } else {
- // in this case, bamFiles and fastqFiles are arrays
+ // in this case fastqFiles are arrays
  // not singular elements per sample
- count_ch.alignments.flatMap { sample ->
- def (sampleName, bamFiles, fastqFiles) = sample
+ // reformat channel for consensus input
+ // to tuple (sampleName, fastqFile)
+ // filter out unmapped and out files from splitcode
+ demux_ch.flatMap { sample ->
+ def (sampleName, fastqFiles) = sample
  return bamFiles.indices.collect { index ->
- [sampleName, bamFiles[index], fastqFiles[index]]
+ [sampleName, fastqFiles[index]]
  }
- }.filter{ sampleName, bamFile, fastqFile ->
- !bamFile.getName().startsWith("unmapped.bam") &&
- !bamFile.getName().startsWith("out.bam")
- }.set{ bam_ch }
+ }.filter{ sampleName, fastqFile ->
+ !fastqFile.getName().startsWith("unmapped.fastq") &&
+ !fastqFile.getName().startsWith("out.fastq")
+ }.set{ fastq_ch }
  }
- Consensus(bam_ch, file(params.guides_fasta), params.medaka_model)
+ Consensus(fastq_ch, file(params.guides_fasta), params.medaka_model)
  }
  }
 }

modules/medaka.nf

@@ -7,15 +7,16 @@ process Medaka {
  publishDir "${params.outdir}/consensus/${sampleName}", mode: 'copy'
 
  conda "${ params.conda_env_location != null && params.conda_env_location != '' ?
- params.conda_env_location + '/racon-medaka' :
- projectDir + '/envs/racon-medaka.yaml' }"
+ params.conda_env_location + '/medaka' :
+ projectDir + '/envs/medaka.yaml' }"
 
  container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
  'https://depot.galaxyproject.org/singularity/medaka:1.11.1--py310h87e71ce_0' :
  'biocontainers/medaka:1.11.1--py310h87e71ce_0' }"
 
  input:
- tuple val(sampleName), path(reads), path(assembly)
+ tuple val(sampleName), path(reads) 
+ path(reference)
  val(medaka_model)
 
  output:
@@ -27,15 +28,15 @@ process Medaka {
  script:
  def args = task.ext.args ?: ''
  def prefix = task.ext.prefix ?: reads.getSimpleName().replaceFirst("rezip_", "")
- // will create a blank fasta file if racon assembly is blank
+ // will create a blank fasta file if fastq file input is blank
  // as for some sample we will not be able to generate a consensus
  """
- if [ -s ${assembly} ]; then
+ if [ -s ${reads} ]; then
  medaka_consensus \\
  -t $task.cpus \\
  $args \\
  -i $reads \\
- -d $assembly \\
+ -d $reference \\
  -m $medaka_model \\
  -o ./
  else

subworkflows/consensus.nf

@@ -1,10 +1,8 @@
-include { Racon } from '../modules/racon'
 include { Medaka } from '../modules/medaka'
 include { AlignPairs } from '../modules/align_pairs.nf'
 
 process PrepareForConsensus {
  label = "PrepareForConsensus"
- // convert bam file to sam for racon and
  // rezip fastq file using bgzip for medaka
 
  publishDir "${params.outdir}/count/${sampleName}"
@@ -14,15 +12,14 @@ process PrepareForConsensus {
  projectDir + '/envs/minimap-samtools.yaml' }"
 
  input:
- tuple val(sampleName), path(bam), path(fastq)
+ tuple val(sampleName), path(fastq)
 
  output:
- tuple val(sampleName), path("*.sam"), path("rezip_*.fastq.gz"), emit: racon_input
+ tuple val(sampleName), path("rezip_*.fastq.gz"), emit: rezipped
 
  script:
- def sample = bam.getSimpleName()
+ def sample = fastq.getSimpleName()
  """
- samtools view -h ${bam} > ${sample}.sam
  zcat < ${fastq} | bgzip -c - > rezip_${sample}.fastq.gz
  """
 }
@@ -36,9 +33,7 @@ workflow Consensus {
  main:
  PrepareForConsensus(consensus_input)
 
- Racon(PrepareForConsensus.out.racon_input, reference)
-
- Medaka(Racon.out.racon_consensus, medaka_model)
+ Medaka(PrepareForConsensus.out.rezipped, reference, medaka_model)
 
  consensus_sequences = Medaka.out.assembly