Add chunk_model as parameter

CHANGELOG.md

@@ -64,6 +64,7 @@ Initial release of nf-core/phaseimpute, created with the [nf-core](https://nf-co
 - [#148](https://github.com/nf-core/phaseimpute/pull/148) - Fix awsfulltest github action for manual dispatch
 - [#149](https://github.com/nf-core/phaseimpute/pull/149) - Remove the map file from the awsfulltest
 - [#152](https://github.com/nf-core/phaseimpute/pull/152) - Fix URLs in the documentation and remove tools citation in the README, use a white background for all images in the documentation.
+- [#157](https://github.com/nf-core/phaseimpute/pull/157) - Add `chunk_model` as parameter for better control over `GLIMPSE2_CHUNK` and set window size in `GLIMPSE1_CHUNK` and `GLIMPSE2_chunk` to 4mb to reduce number of chunks (empirical).
 
 ### `Fixed`
 

conf/steps/panel_prep.config

@@ -41,8 +41,8 @@ process {
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_NORMALIZE_BCFTOOLS:BCFTOOLS_NORM' {
-        ext.args      = ["-m +any", "--no-version", "--output-type z", "--write-index=tbi"].join(' ')
-        ext.prefix    = { "${meta.id}_${meta.chr}_multiallelic" }
+        ext.args   = ["-m +any", "--no-version", "--output-type z", "--write-index=tbi"].join(' ')
+        ext.prefix = { "${meta.id}_${meta.chr}_multiallelic" }
         publishDir = [ enabled: false ]
     }
 
@@ -62,7 +62,7 @@ process {
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_NORMALIZE_BCFTOOLS:VCFLIB_VCFFIXUP' {
-        ext.prefix   = { "${meta.id}_${meta.chr}_fixup" }
+        ext.prefix = { "${meta.id}_${meta.chr}_fixup" }
         publishDir = [
             path: { "${params.outdir}/prep_panel/panel" },
             mode: params.publish_dir_mode,
@@ -72,7 +72,7 @@ process {
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_NORMALIZE_BCFTOOLS:BCFTOOLS_INDEX' {
-        ext.args     = "--tbi"
+        ext.args   = "--tbi"
         publishDir = [
             path: { "${params.outdir}/prep_panel/panel" },
             mode: params.publish_dir_mode,
@@ -88,6 +88,7 @@ process {
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_PHASE_SHAPEIT5:GLIMPSE2_CHUNK' {
+        ext.args   = "--window-mb 4"
         ext.prefix = { "${meta.id}_chunks" }
     }
 
@@ -123,7 +124,7 @@ process {
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_SITES_EXTRACT_BCFTOOLS:BCFTOOLS_CONVERT' {
-        ext.args = {"--haplegendsample ${meta.id}_${meta.chr}"}
+        ext.args   = {"--haplegendsample ${meta.id}_${meta.chr}"}
         publishDir = [
             path: { "${params.outdir}/prep_panel/haplegend/" },
             mode: params.publish_dir_mode,
@@ -184,6 +185,7 @@ process {
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_CHUNK_GLIMPSE:GLIMPSE_CHUNK' {
+        ext.args   = "--window-size 4"
         ext.prefix = { "${meta.id}_${meta.chr}_chunks_glimpse1" }
         publishDir = [
             path: { "${params.outdir}/prep_panel/chunks/glimpse1/" },
@@ -194,7 +196,8 @@ process {
     }
 
     withName: 'NFCORE_PHASEIMPUTE:PHASEIMPUTE:VCF_CHUNK_GLIMPSE:GLIMPSE2_CHUNK' {
-        ext.prefix    = { "${meta.id}_${meta.chr}_chunks_glimpse2" }
+        ext.args   = "--window-mb 4"
+        ext.prefix = { "${meta.id}_${meta.chr}_chunks_glimpse2" }
         publishDir = [
             path: { "${params.outdir}/prep_panel/chunks/glimpse2/" },
             mode: params.publish_dir_mode,

conf/test_all.config

@@ -33,6 +33,7 @@ params {
     phase        = true
     normalize    = true
     compute_freq = false
+    chunk_model  = "recursive"
 
     // Pipeline steps
     steps = "all"

conf/test_dog.config

@@ -32,6 +32,7 @@ params {
     normalize    = false
     compute_freq = false
     rename_chr   = true
+    chunk_model  = "recursive"
 
     // Input data
     input        = params.pipelines_testdata_base_path + "dog_data/csv/sample_dog.csv"

conf/test_panelprep.config

@@ -32,6 +32,7 @@ params {
     normalize      = true
     compute_freq   = true
     remove_samples = "HG00096,HG00097,HG00099,HG00100"
+    chunk_model    = "recursive"
 
     // Pipeline steps
     steps   = "panelprep"

nextflow.config

@@ -10,7 +10,7 @@
 params {
 
     // steps
-    steps                        = null
+    steps                       = null
 
     // Input options
     input                       = null
@@ -24,6 +24,7 @@ params {
     normalize                   = true
     compute_freq                = false
     remove_samples              = null
+    chunk_model                 = 'sequential'
 
     // ChrCheck parameters
     rename_chr                  = false

nextflow_schema.json

@@ -132,6 +132,13 @@
                 "binaryref": {
                     "type": "string",
                     "description": "Whether to generate a binary reference file to be used with GLIMPSE2"
+                },
+                "chunk_model": {
+                    "type": "string",
+                    "description": "Model type to use for GLIMPSE2_CHUNK",
+                    "enum": ["recursive", "sequential", "uniform-number-variants"],
+                    "default": "sequential",
+                    "hidden": true
                 }
             }
         },

subworkflows/local/vcf_chunk_glimpse/main.nf

@@ -7,6 +7,7 @@ workflow VCF_CHUNK_GLIMPSE {
     take:
     ch_reference  // channel: [ [panel, chr], vcf, csi ]
     ch_map        // channel  (optional): [ [chr], map ]
+    chunk_model   // channel : model
 
     main:
 
@@ -36,9 +37,6 @@ workflow VCF_CHUNK_GLIMPSE {
         )
         .map { metaPC, it -> [metaPC, it["RegionIn"], it["RegionOut"]]}
 
-    // Make chunks with Glimpse2 (does not work with "sequential" mode)
-    chunk_model = "recursive"
-
     ch_input_glimpse2 = ch_vcf_csi_chr
         .map{
             metaPC, vcf, csi, chr -> [metaPC.subMap("chr"), metaPC, vcf, csi, chr]

subworkflows/local/vcf_chunk_glimpse/tests/main.nf.test

@@ -47,6 +47,7 @@ nextflow_workflow {
                         file(params.pipelines_testdata_base_path + "hum_data/reference_genome/GRCh38_21.map", checkIfExist:true)
                     ]
                 )
+                input[2] = "recursive"
                 """
             }
         }
@@ -87,6 +88,7 @@ nextflow_workflow {
                     [[chr: "chr22"], []],
                     [[chr: "chr21"], []]
                 )
+                input[2] = "recursive"
                 """
             }
         }

subworkflows/local/vcf_phase_shapeit5/main.nf

@@ -7,19 +7,17 @@ include { BCFTOOLS_INDEX as VCF_BCFTOOLS_INDEX_2 } from '../../../modules/nf-cor
 workflow VCF_PHASE_SHAPEIT5 {
 
     take:
-    ch_vcf        // channel (mandatory): [ [id, chr], vcf, csi, pedigree ]
+    ch_vcf        // channel (mandatory) : [ [id, chr], vcf, csi, pedigree ]
     ch_region     // channel (mandatory) : [ [chr, region], region ]
-    ch_ref        // channel (optional) : [ [id, chr], ref, csi ]
-    ch_scaffold   // channel (optional) : [ [id, chr], scaffold, csi ]
+    ch_ref        // channel (optional)  : [ [id, chr], ref, csi ]
+    ch_scaffold   // channel (optional)  : [ [id, chr], scaffold, csi ]
     ch_map        // channel (mandatory) : [ [chr], map]
+    chunk_model   // channel (mandatory) : [ model ]
 
     main:
 
     ch_versions = Channel.empty()
 
-    // Make chunks with Glimpse2 (does not work with "sequential" mode)
-    chunk_model = "recursive"
-
     // Chunk with Glimpse2
     ch_input_glimpse2 = ch_vcf
         .map{

subworkflows/local/vcf_phase_shapeit5/tests/main.nf.test

@@ -52,6 +52,7 @@ nextflow_workflow {
                     [[chr: "chr22"],[]],
                     [[chr: "chr21"], []]
                 )
+                input[5] = "recursive"
                 """
             }
         }
@@ -103,6 +104,7 @@ nextflow_workflow {
                     [ [chr: "chr22"], file(params.pipelines_testdata_base_path + "hum_data/reference_genome/GRCh38_22.map", checkIfExist:true)],
                     [ [chr: "chr21"], file(params.pipelines_testdata_base_path + "hum_data/reference_genome/GRCh38_21.map", checkIfExist:true)]
                 )
+                input[5] = "recursive"
                 """
             }
         }

workflows/phaseimpute/main.nf

@@ -171,14 +171,15 @@ workflow PHASEIMPUTE {
                 ch_region,
                 [[],[],[]],
                 [[],[],[]],
-                ch_map
+                ch_map,
+                params.chunk_model
             )
             ch_panel_phased = VCF_PHASE_SHAPEIT5.out.vcf_tbi
             ch_versions = ch_versions.mix(VCF_PHASE_SHAPEIT5.out.versions)
         }
 
         // Create chunks from reference VCF
-        VCF_CHUNK_GLIMPSE(ch_panel_phased, ch_map)
+        VCF_CHUNK_GLIMPSE(ch_panel_phased, ch_map, params.chunk_model)
         ch_versions = ch_versions.mix(VCF_CHUNK_GLIMPSE.out.versions)
 
         // Assign chunks channels