Run a single phasing software instead of three in parallel (genomic-m…

…edicine-sweden#83)
fellen31 · Apr 18, 2024 · 7fdb5fd · 7fdb5fd
1 parent d293da8
commit 7fdb5fd
Show file tree

Hide file tree

Showing 15 changed files with 355 additions and 254 deletions.
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
@@ -25,7 +25,7 @@ jobs:
       matrix:
         parameters:
           - ""
-          - "--input assets/samplesheet_multisample_bam.csv --split_fastq 250 --parallel_snv 1"
+          - "--input assets/samplesheet_multisample_bam.csv --split_fastq 250 --parallel_snv 1 --phaser hiphase_sv"
         NXF_VER:
           - "23.04.0"
           - "latest-everything"

diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -9,6 +9,7 @@ Initial release of genomic-medicine-sweden/skierfe, created with the [nf-core](h
 
 ### `Added`
 
+- Added a switch to choose phasing software [#83](https://github.com/genomic-medicine-sweden/skierfe/pull/83)
 - Added SNV annotation subworkflow to test profile [#75](https://github.com/genomic-medicine-sweden/skierfe/pull/75)
 - Refactored conditionally required parameters validation [#69](https://github.com/genomic-medicine-sweden/skierfe/pull/69)
 - Added CNV workflow to test profile [#68](https://github.com/genomic-medicine-sweden/skierfe/pull/68)

diff --git a/conf/modules/phasing.config b/conf/modules/phasing.config
@@ -26,56 +26,95 @@ process {
 
     withName: '.*:PHASING:HIPHASE_SNV' {
 
-        ext.args = "--ignore-read-groups"
+        ext.args = { [
+            '--ignore-read-groups',
+            "--stats-file ${meta.id}.stats.tsv",
+            "--blocks-file ${meta.id}.blocks.tsv",
+            "--summary-file ${meta.id}.summary.tsv"
+        ].join(' ') }
 
         publishDir = [
-            path: { "${params.outdir}/phased_reads/hiphase/snv/${meta.id}" },
+            path: { "${params.outdir}/phasing/hiphase/snv/${meta.id}" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
     }
 
     withName: '.*:PHASING:HIPHASE_SV' {
 
-        ext.args = "--ignore-read-groups --global-realignment-cputime 300"
+        ext.args = { [
+            '--ignore-read-groups',
+            "--stats-file ${meta.id}.stats.tsv",
+            "--blocks-file ${meta.id}.blocks.tsv",
+            "--summary-file ${meta.id}.summary.tsv"
+        ].join(' ') }
 
         publishDir = [
-            path: { "${params.outdir}/phased_reads/hiphase/sv/${meta.id}" },
+            path: { "${params.outdir}/phasing/hiphase/sv/${meta.id}" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
     }
 
     withName: '.*:PHASING:WHATSHAP_PHASE' {
-        // Do not use phased variant calls for analysis - so don't publish
-        ext.args = '--ignore-read-groups --indels --distrust-genotypes'
+
+        ext.args = [
+            '--ignore-read-groups',
+            '--indels'
+        ].join(' ')
+
+        publishDir = [
+            path: { "${params.outdir}/phasing/whatshap/phase/${meta.id}" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
     }
 
     withName: '.*:PHASING:WHATSHAP_STATS' {
+
         publishDir = [
-            path: { "${params.outdir}/phased_reads/whatshap/stats/${meta.id}" },
+            path: { "${params.outdir}/phasing/whatshap/stats/${meta.id}" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
     }
 
     withName: '.*:PHASING:WHATSHAP_HAPLOTAG' {
-        // TODO: Should supplementary be tagged?
-        ext.args = '--ignore-read-groups --tag-supplementary'
+
+        ext.args = [
+            '--ignore-read-groups',
+            '--tag-supplementary'
+        ].join(' ')
 
         publishDir = [
-            path: { "${params.outdir}/phased_reads/whatshap/haplotag/${meta.id}" },
+            path: { "${params.outdir}/phasing/whatshap/haplotag/${meta.id}" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
     }
 
     withName: '.*:PHASING:SAMTOOLS_INDEX_WHATSHAP' {
+
         publishDir = [
-            path: { "${params.outdir}/phased_reads/whatshap/haplotag/${meta.id}" },
+            path: { "${params.outdir}/phasing/whatshap/haplotag/${meta.id}" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
     }
 
+    withName: '.*:PHASING:CRAMINO_PHASED' {
+
+        ext.args = [
+            '--karyotype',
+            '--phased'
+        ].join(' ')
+
+        publishDir = [
+            path: { "${params.outdir}/phasing/cramino/${meta.id}" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+
+    }
+
 }
diff --git a/conf/modules/qc.config b/conf/modules/qc.config
@@ -27,16 +27,6 @@ process {
         ]
     }
 
-    withName: '.*:QC_ALIGNED_READS:CRAMINO_PHASED' {
-        ext.args = '--karyotype --phased'
-        publishDir = [
-            path: { "${params.outdir}/qc_aligned_reads/cramino/phased/${meta.id}" },
-            mode: params.publish_dir_mode,
-            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
-        ]
-
-    }
-
     withName: '.*:QC_ALIGNED_READS:MOSDEPTH' {
 
         // If pipeline is run with bed-file, then don't run wth '--by 500'

diff --git a/docs/output.md b/docs/output.md
@@ -23,10 +23,12 @@ This document roughly describes the output structure produced by the pipeline. T
 | multiqc                   | Directory for MultiQC reports and plots.                                      |
 | ├── multiqc_data          | Directory containing data files generated by MultiQC.                         |
 | └── multiqc_plots         | Directory containing plots generated by MultiQC.                              |
-| phased_reads              | Directory for phased read analysis.                                           |
+| phasing                   | Directory for phasing analysis.                                               |
+| ├── cramino               | Directory containing QC results for phased reads using the cramino tool.      |
 | ├── hiphase               | Directory containing phased SNV and SV analysis results by Hiphase.           |
 | └── whatshap              | Directory containing phased reads and statistics by WhatsHap.                 |
 | &emsp;├── haplotag        | Directory containing haplotagging results.                                    |
+| &emsp;├── phase           | Directory containing phased variants.                                         |
 | &emsp;└── stats           | Directory containing statistics related to phased reads.                      |
 | pipeline_info             | Directory containing information and reports about the pipeline.              |
 | qc_aligned_reads          | Directory for quality control results of aligned reads.                       |