From 10ca39a86f1b227a5535238d02398fec686eba72 Mon Sep 17 00:00:00 2001
From: "Maxime U. Garcia" <max.u.garcia@gmail.com>
Date: Wed, 27 Apr 2022 16:03:44 +0200
Subject: [PATCH] add intervals possibilities to splitncigarreads (#1571)

---
 modules/gatk4/splitncigarreads/main.nf        |  4 +++-
 modules/gatk4/splitncigarreads/meta.yml       |  7 +++++++
 tests/modules/gatk4/splitncigarreads/main.nf  | 18 +++++++++++++++++-
 tests/modules/gatk4/splitncigarreads/test.yml | 11 ++++++++++-
 4 files changed, 37 insertions(+), 3 deletions(-)

diff --git a/modules/gatk4/splitncigarreads/main.nf b/modules/gatk4/splitncigarreads/main.nf
index f7c559d9..85e5daa8 100644
--- a/modules/gatk4/splitncigarreads/main.nf
+++ b/modules/gatk4/splitncigarreads/main.nf
@@ -8,7 +8,7 @@ process GATK4_SPLITNCIGARREADS {
         'quay.io/biocontainers/gatk4:4.2.5.0--hdfd78af_0' }"
 
     input:
-    tuple val(meta), path(bam)
+    tuple val(meta), path(bam), path(bai), path(intervals)
     path  fasta
     path  fai
     path  dict
@@ -23,6 +23,7 @@ process GATK4_SPLITNCIGARREADS {
     script:
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
+    def interval_command = intervals ? "--intervals $intervals" : ""
 
     def avail_mem = 3
     if (!task.memory) {
@@ -35,6 +36,7 @@ process GATK4_SPLITNCIGARREADS {
         --input $bam \\
         --output ${prefix}.bam \\
         --reference $fasta \\
+        $interval_command \\
         --tmp-dir . \\
         $args
 
diff --git a/modules/gatk4/splitncigarreads/meta.yml b/modules/gatk4/splitncigarreads/meta.yml
index 407e80bd..76bfdcd3 100644
--- a/modules/gatk4/splitncigarreads/meta.yml
+++ b/modules/gatk4/splitncigarreads/meta.yml
@@ -23,6 +23,13 @@ input:
       type: list
       description: BAM/SAM/CRAM file containing reads
       pattern: "*.{bam,sam,cram}"
+  - bai:
+      type: list
+      description: BAI/SAI/CRAI index file (optional)
+      pattern: "*.{bai,sai,crai}"
+  - intervals:
+      type: file
+      description: Bed file with the genomic regions included in the library (optional)
   - fasta:
       type: file
       description: The reference fasta file
diff --git a/tests/modules/gatk4/splitncigarreads/main.nf b/tests/modules/gatk4/splitncigarreads/main.nf
index 7e5b7c9a..31e45cec 100644
--- a/tests/modules/gatk4/splitncigarreads/main.nf
+++ b/tests/modules/gatk4/splitncigarreads/main.nf
@@ -6,7 +6,23 @@ include { GATK4_SPLITNCIGARREADS } from '../../../../modules/gatk4/splitncigarre
 
 workflow test_gatk4_splitncigarreads {
     input = [ [ id:'test' ], // meta map
-              [ file(params.test_data['sarscov2']['illumina']['test_paired_end_bam'], checkIfExists: true) ] 
+                file(params.test_data['sarscov2']['illumina']['test_paired_end_bam'], checkIfExists: true),
+                [],
+                []
+            ]
+
+    fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    fai = file(params.test_data['sarscov2']['genome']['genome_fasta_fai'], checkIfExists: true)
+    dict = file(params.test_data['sarscov2']['genome']['genome_dict'], checkIfExists: true)
+
+    GATK4_SPLITNCIGARREADS ( input, fasta, fai, dict )
+}
+
+workflow test_gatk4_splitncigarreads_intervals {
+    input = [ [ id:'test' ], // meta map
+                file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
+                file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true),
+                file(params.test_data['sarscov2']['genome']['test_bed'], checkIfExists: true)
             ]
     fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
     fai = file(params.test_data['sarscov2']['genome']['genome_fasta_fai'], checkIfExists: true)
diff --git a/tests/modules/gatk4/splitncigarreads/test.yml b/tests/modules/gatk4/splitncigarreads/test.yml
index c38064e2..d9a58901 100644
--- a/tests/modules/gatk4/splitncigarreads/test.yml
+++ b/tests/modules/gatk4/splitncigarreads/test.yml
@@ -5,5 +5,14 @@
     - gatk4/splitncigarreads
   files:
     - path: output/gatk4/test.bam
-      md5sum: ceed15c0bd64ff5c38d3816905933b0b
+      md5sum: 436d8e31285c6b588bdd1c7f1d07f6f2
+    - path: output/gatk4/versions.yml
+- name: gatk4 splitncigarreads test_gatk4_splitncigarreads_intervals
+  command: nextflow run tests/modules/gatk4/splitncigarreads -entry test_gatk4_splitncigarreads_intervals -c tests/config/nextflow.config
+  tags:
+    - gatk4
+    - gatk4/splitncigarreads
+  files:
+    - path: output/gatk4/test.bam
+      md5sum: cd56e3225950f519fd47164cca60a0bb
     - path: output/gatk4/versions.yml