Merge pull request #448 from FriederikeHanssen/skip_bqsr

allow tp skip bqsr

Author: FriederikeHanssen

CHANGELOG.md

@@ -9,6 +9,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Added
 
+- [##448](https://github.com/nf-core/sarek/pull/448) - Allow to skip base quality recalibration with `--skip_bqsr`
+
 ### Changed
 
 ### Fixed

nextflow.config

@@ -47,6 +47,9 @@ params {
     skip_markduplicates = false // Do not skip markDuplicates by default
     sequencing_center = null // No sequencing center to be written in BAM header in MapReads process
 
+    //BQSR
+    skip_bqsr = false
+
     // Variant Calling
     ascat_ploidy = null // Use default value
     ascat_purity = null // Use default value

nextflow_schema.json

@@ -219,6 +219,12 @@
                     "fa_icon": "fas fa-fast-forward",
                     "description": "Skip GATK MarkDuplicates",
                     "help_text": "This params will also save the mapped BAMS, to enable restart from step `prepare_recalibration`"
+                },
+                "skip_bqsr": {
+                    "type": "boolean",
+                    "default": false,
+                    "fa_icon": "fas fa-fast-forward",
+                    "description": "Skip Base Quality Recalibration"
                 }
             }
         },

workflows/sarek.nf

@@ -383,48 +383,56 @@ workflow SAREK {
         qc_reports = qc_reports.mix(MARKDUPLICATES.out.qc.collect{it[1]}.ifEmpty([]))
 
         // STEP 3: Create recalibration tables
-        PREPARE_RECALIBRATION(
-            cram_markduplicates,
-            ('bqsr' in params.use_gatk_spark),
-            dict,
-            fai,
-            fasta,
-            intervals,
-            num_intervals,
-            known_sites,
-            known_sites_tbi,
-            params.no_intervals)
-
-        table_bqsr = PREPARE_RECALIBRATION.out.table_bqsr
-        PREPARE_RECALIBRATION_CSV(table_bqsr)
-
-        cram_applybqsr = cram_markduplicates.join(table_bqsr)
+        if(!params.skip_bqsr){
+            PREPARE_RECALIBRATION(
+                cram_markduplicates,
+                ('bqsr' in params.use_gatk_spark),
+                dict,
+                fai,
+                fasta,
+                intervals,
+                num_intervals,
+                known_sites,
+                known_sites_tbi,
+                params.no_intervals)
+
+            table_bqsr = PREPARE_RECALIBRATION.out.table_bqsr
+            PREPARE_RECALIBRATION_CSV(table_bqsr)
+
+            cram_applybqsr = cram_markduplicates.join(table_bqsr)
+        }
     }
 
     if (step == 'recalibrate') bam_applybqsr = input_sample
 
     if (step in ['mapping', 'preparerecalibration', 'recalibrate']) {
-        // STEP 4: RECALIBRATING
-        RECALIBRATE(
-            ('bqsr' in params.use_gatk_spark),
-            ('bamqc' in skip_qc),
-            ('samtools' in skip_qc),
-            cram_applybqsr,
-            dict,
-            fai,
-            fasta,
-            intervals,
-            num_intervals,
-            target_bed)
-
-        cram_recalibrated    = RECALIBRATE.out.cram
-        cram_recalibrated_qc = RECALIBRATE.out.qc
 
-        RECALIBRATE_CSV(cram_recalibrated)
-
-        qc_reports = qc_reports.mix(cram_recalibrated_qc.collect{it[1]}.ifEmpty([]))
+        if(!params.skip_bqsr){
+            // STEP 4: RECALIBRATING
+            RECALIBRATE(
+                ('bqsr' in params.use_gatk_spark),
+                ('bamqc' in skip_qc),
+                ('samtools' in skip_qc),
+                cram_applybqsr,
+                dict,
+                fai,
+                fasta,
+                intervals,
+                num_intervals,
+                target_bed)
+
+            cram_recalibrated    = RECALIBRATE.out.cram
+            cram_recalibrated_qc = RECALIBRATE.out.qc
+
+            RECALIBRATE_CSV(cram_recalibrated)
+
+            qc_reports = qc_reports.mix(cram_recalibrated_qc.collect{it[1]}.ifEmpty([]))
+            cram_variant_calling = cram_recalibrated
+
+        }else{
+            cram_variant_calling = cram_markduplicates
+        }
 
-        cram_variant_calling = cram_recalibrated
     }
 
     if (step in 'variantcalling') cram_variant_calling = input_sample