fastp to cellranger

nf-core · May 7, 2024 · b092a25 · b092a25
1 parent 2f492b0
commit b092a25
Show file tree

Hide file tree

Showing 3 changed files with 53 additions and 3 deletions.
diff --git a/modules/local/rename_fastq_cellranger.nf b/modules/local/rename_fastq_cellranger.nf
@@ -0,0 +1,23 @@
+// Import generic module functions
+process RENAME_FASTQ_CELLRANGER {
+    tag "$meta.id"
+    label 'process_low'
+
+    conda "conda-forge::python=3.8.0 conda-forge::biopython=1.74"
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/mulled-v2-adc9bb9edc31eb38b3c24786a83b7dfa530e2bea:47d6d7765d7537847ced7dac873190d164146022-0' :
+        'biocontainers/mulled-v2-adc9bb9edc31eb38b3c24786a83b7dfa530e2bea:47d6d7765d7537847ced7dac873190d164146022-0' }"
+
+    input:
+    tuple val(meta), path(R1), path(R2)
+    tuple val(meta_2), path(orig_r1), path(orig_r2)
+
+    output:
+    tuple val(meta), path('*R1_001.fastq.gz'), path('*R2_001.fastq.gz') , emit: reads
+
+    script:
+    """
+    mv ${R1} fastp_${orig_r1}
+    mv ${R2} fastp_${orig_r2}
+    """
+}
diff --git a/subworkflows/local/sc_raw_input.nf b/subworkflows/local/sc_raw_input.nf
@@ -3,6 +3,9 @@ include { UNZIP_CELLRANGERDB                                            } from '
 include { RENAME_FILE as RENAME_FILE_TSV                                } from '../../modules/local/rename_file'
 include { CHANGEO_CONVERTDB_FASTA as CHANGEO_CONVERTDB_FASTA_FROM_AIRR  } from '../../modules/local/changeo/changeo_convertdb_fasta'
 include { FASTQ_INPUT_CHECK                                             } from '../../subworkflows/local/fastq_input_check'
+include { FASTP                                                         } from '../../modules/nf-core/fastp/main'
+include { RENAME_FASTQ_CELLRANGER                                       } from '../../modules/local/rename_fastq_cellranger'
+
 
 
 workflow SC_RAW_INPUT {
@@ -51,9 +54,30 @@ workflow SC_RAW_INPUT {
         error "The single-cell 10X genomics library generation method requires you to provide a reference file."
     }
 
+    // Fastp
+    save_merged = false
+    FASTP (
+        ch_reads,
+        [],
+        [],
+        save_merged
+    )
+    ch_versions = ch_versions.mix(FASTP.out.versions)
+
+    ch_rename_fastp = FASTP.out.reads.map{ meta,reads -> [meta, reads[0], reads[1]] }
+    ch_rename_original = ch_reads.map{ meta,reads -> [meta, reads[0], reads[1]] }
+
+    // rename fastq files to follow cellranger standards again
+    RENAME_FASTQ_CELLRANGER(
+        ch_rename_fastp,
+        ch_rename_original
+    )
+
+    ch_reads_fastp = RENAME_FASTQ_CELLRANGER.out.reads.map{ meta, read1, read2 -> [meta, [read1, read2]] }
+
     // run cellranger vdj
     CELLRANGER_VDJ (
-        ch_reads,
+        ch_reads_fastp,
         ch_sc_reference.collect()
     )
     ch_versions = ch_versions.mix(CELLRANGER_VDJ.out.versions)
@@ -89,6 +113,9 @@ workflow SC_RAW_INPUT {
 
     emit:
     versions = ch_versions
+    // fastp
+    fastp_reads_json = FASTP.out.json.collect{ meta,json -> json }
+    fastp_reads_html = FASTP.out.html.collect{ meta,html -> html }
     // complete cellranger output
     outs = ch_cellranger_out
     // cellranger output in airr format

diff --git a/workflows/airrflow.nf b/workflows/airrflow.nf
@@ -103,8 +103,8 @@ workflow AIRRFLOW {
                 ch_presto_assemblepairs_logs            = Channel.empty()
                 ch_presto_collapseseq_logs              = Channel.empty()
                 ch_presto_splitseq_logs                 = Channel.empty()
-                ch_fastp_html                           = Channel.empty()
-                ch_fastp_json                           = Channel.empty()
+                ch_fastp_html                           = SC_RAW_INPUT.out.fastp_reads_html
+                ch_fastp_json                           = SC_RAW_INPUT.out.fastp_reads_json
                 ch_fastqc_postassembly_mqc              = Channel.empty()
             } else {
                 // Perform sequence assembly if input type is fastq from bulk sequencing data