add step to transform BED file into intervals

TRON-Bioinformatics · Nov 21, 2021 · 76368ed · 76368ed
1 parent 233ab0b
commit 76368ed
Show file tree

Hide file tree

Showing 3 changed files with 10 additions and 8 deletions.
diff --git a/main.nf b/main.nf
@@ -17,7 +17,6 @@ params.dbsnp = false
 params.known_indels1 = false
 params.known_indels2 = false
 params.intervals = false
-params.intervals_bed = false
 params.skip_bqsr = false
 params.skip_realignment = false
 params.skip_deduplication = false

diff --git a/modules/03_metrics.nf b/modules/03_metrics.nf
@@ -4,7 +4,6 @@ params.collect_hs_metrics_min_base_quality = false
 params.collect_hs_metrics_min_mapping_quality = false
 params.reference = false
 params.output = 'output'
-params.intervals_bed = false
 params.intervals = false
 
 
@@ -32,12 +31,17 @@ process HS_METRICS {
     """
     mkdir tmp
 
+    gatk BedToIntervalList \
+    --INPUT ${params.intervals} \
+    --OUTPUT ${params.intervals}.intervals \
+    --SEQUENCE_DICTIONARY ${bam}
+
     gatk CollectHsMetrics \
     --java-options '-Xmx${params.metrics_memory}  -Djava.io.tmpdir=tmp' \
     --INPUT  ${bam} \
     --OUTPUT ${name}.hs_metrics.txt \
-    --TARGET_INTERVALS ${params.intervals} \
-    --BAIT_INTERVALS ${params.intervals} \
+    --TARGET_INTERVALS ${params.intervals}.intervals \
+    --BAIT_INTERVALS ${params.intervals}.intervals \
     ${minimum_base_quality} ${minimum_mapping_quality}
     """
 }
@@ -96,7 +100,7 @@ process COVERAGE_ANALYSIS {
         "--min-BQ ${params.collect_hs_metrics_min_base_quality}" : ""
     minimum_mapping_quality = params.collect_hs_metrics_min_mapping_quality ?
         "--min-MQ ${params.collect_hs_metrics_min_mapping_quality}" : ""
-    intervals = params.intervals_bed ? "-b ${params.intervals_bed}" : ""
+    intervals = params.intervals ? "-b ${params.intervals}" : ""
     """
     samtools coverage ${minimum_base_quality} ${minimum_mapping_quality} ${bam} > ${name}.coverage.tsv
     samtools depth -s -d 0 -H ${intervals} ${bam} > ${name}.depth.tsv

diff --git a/nextflow.config b/nextflow.config
@@ -27,8 +27,7 @@ profiles {
     params.index_memory = "3g"
     params.known_indels1 = "$baseDir/test_data/1000G_phase1.indels.hg19.sites.minimal.vcf"
     params.known_indels2 = "$baseDir/test_data/Mills_and_1000G_gold_standard.indels.hg19.sites.sorted.minimal.vcf"
-    params.intervals = "$baseDir/test_data/minimal_intervals.intervals"
-    params.intervals_bed = "$baseDir/test_data/minimal_intervals.bed"
+    params.intervals = "$baseDir/test_data/minimal_intervals.bed"
     params.dbsnp = "$baseDir/test_data/dbsnp_138.hg19.minimal.vcf"
     timeline.enabled = false
     report.enabled = false
@@ -83,7 +82,7 @@ Optional input:
     * --dbsnp: path to the dbSNP VCF (required to perform BQSR)
     * --known_indels1: path to a VCF of known indels (optional to perform realignment around indels)
     * --known_indels2: path to a second VCF of known indels (optional to perform realignment around indels)
-    * --intervals: path to an intervals file to collect HS metrics from, this can be built with Picard's BedToIntervalList (default: None)
+    * --intervals: path to a BED file to collect coverage and HS metrics from (default: None)
     * --collect_hs_minimum_base_quality: minimum base quality for a base to contribute coverage (default: 20).
     * --collect_hs_minimum_mapping_quality: minimum mapping quality for a read to contribute coverage (default: 20).
     * --skip_bqsr: optionally skip BQSR (default: false)