From 1677b34514424a5f38716f42daaa862e698d2e35 Mon Sep 17 00:00:00 2001
From: Sebastian Schoenherr <sebastian.schoenherr@i-med.ac.at>
Date: Fri, 14 Jun 2024 14:38:16 +0200
Subject: [PATCH] Add phasing window (#16)

* Add phasing window

* Add phasing window to beagle

* Simplify syntax

* Fix typo

* Remove characters in beagle call
---
 modules/local/imputation/minimac4.nf |  8 ++++----
 modules/local/phasing/beagle.nf      | 13 ++++++++-----
 modules/local/phasing/eagle.nf       | 14 ++++++++------
 3 files changed, 20 insertions(+), 15 deletions(-)

diff --git a/modules/local/imputation/minimac4.nf b/modules/local/imputation/minimac4.nf
index bee938f..825d441 100644
--- a/modules/local/imputation/minimac4.nf
+++ b/modules/local/imputation/minimac4.nf
@@ -11,10 +11,10 @@ process MINIMAC4 {
 
     script:
     def map = minimac_map ? '--referenceEstimates --map ' + minimac_map : ''
-    def chunkfile_name = "${chunkfile}".replaceAll('.vcf.gz', '')
-    def chr_cleaned = "${chr}".startsWith('X.') ? 'X' : "${chr}"
-    def chr_mapped = "${params.refpanel.build}" == 'hg38' ? 'chr' + "${chr_cleaned}" : "${chr_cleaned}"
-    def isChrM = "${chr}" == 'MT' ? '--myChromosome ' + "${chr}" : ''
+    def chunkfile_name = chunkfile.toString().replaceAll('.vcf.gz', '')
+    def chr_cleaned = chr.startsWith('X.') ? 'X' : chr
+    def chr_mapped = (params.refpanel.build == 'hg38') ? 'chr' + chr_cleaned : chr_cleaned
+    def isChrM = (chr == 'MT') ? '--myChromosome ' + chr : ''
 
     """
     minimac4 \
diff --git a/modules/local/phasing/beagle.nf b/modules/local/phasing/beagle.nf
index 8dc6821..ba545a7 100644
--- a/modules/local/phasing/beagle.nf
+++ b/modules/local/phasing/beagle.nf
@@ -10,18 +10,21 @@ process BEAGLE {
 
     script:
     //define basename without ending (do not use simpleName due to X.*)
-    def chunkfile_name = "$chunkfile".replaceAll('.vcf.gz', '')
+    def chunkfile_name = chunkfile.toString().replaceAll('.vcf.gz', '')
     // replace X.nonPAR etc with X for phasing
-    def chr_cleaned = "${chr}".startsWith('X.') ? 'X' : "${chr}"
-    def chr_mapped = "${params.refpanel.build}" == 'hg38' ? 'chr' + "${chr_cleaned}" : "${chr_cleaned}"
-  
+    def chr_cleaned = chr.startsWith('X.') ? 'X' : chr
+    def chr_mapped = params.refpanel.build == 'hg38' ? 'chr' + chr_cleaned : chr_cleaned
+    def phasing_start = start.toLong() - params.phasing_window
+    phasing_start = phasing_start < 0 ? 1 : phasing_start
+    def phasing_end = end.toLong() + params.phasing_window
+    
     """
     java -jar /usr/bin/beagle.18May20.d20.jar \
         ref=${bcf}  \
         gt=${chunkfile} \
         out=${chunkfile_name}.phased \
         nthreads=${task.cpus} \
-        chrom=${chr_mapped}:${start}-${end} \
+        chrom=${chr_mapped}:${phasing_start}-${phasing_end} \
         map=${map_beagle} \
         impute=false 
     """
diff --git a/modules/local/phasing/eagle.nf b/modules/local/phasing/eagle.nf
index d8c6692..b3a197b 100644
--- a/modules/local/phasing/eagle.nf
+++ b/modules/local/phasing/eagle.nf
@@ -11,11 +11,13 @@ process EAGLE {
 
     script:
     //define basename without ending (do not use simpleName due to X.*)
-    def chunkfile_name = "$chunkfile".replaceAll('.vcf.gz', '')
+    def chunkfile_name = chunkfile.toString().replaceAll('.vcf.gz', '')
     // replace X.nonPAR etc with X for phasing
-    def chr_cleaned = "${chr}".startsWith('X.') ? 'X' : "${chr}"
-    def chr_mapped = "${params.refpanel.build}" == 'hg38' ? 'chr' + "${chr_cleaned}" : "${chr_cleaned}"
- 
+    def chr_cleaned = chr.startsWith('X.') ? 'X' : chr
+    def chr_mapped = params.refpanel.build == 'hg38' ? 'chr' + chr_cleaned : chr_cleaned
+    def phasing_start = start.toLong() - params.phasing_window
+    phasing_start = phasing_start < 0 ? 1 : phasing_start
+    def phasing_end = end.toLong() + params.phasing_window
     """
     tabix $chunkfile
     eagle \
@@ -24,8 +26,8 @@ process EAGLE {
         --geneticMapFile ${map_eagle} \
         --outPrefix ${chunkfile_name}.phased \
         --chrom $chr_mapped \
-        --bpStart $start \
-        --bpEnd $end \
+        --bpStart $phasing_start \
+        --bpEnd $phasing_end \
         --allowRefAltSwap \
         --vcfOutFormat z \
         --keepMissingPloidyX \