Add HUMID module for UMI deduplication in nf-core/humid

Joon-Klaps · Joon-Klaps · Apr 30, 2024 · Jan 31, 2024 · Jan 31, 2024 · Jan 31, 2024
commit 88053902f34362dd37543ee2aa6315355b27545b
diff --git a/conf/tests/test_umi.config b/conf/tests/test_umi.config
@@ -30,8 +30,9 @@ params {
     fastp_deduplicate           = true
     fastp_dedup_accuracy        = 2
     skip_umi_extract            = false
+    umi_deduplicate             = 'both'
     host_k2_db                  = 'https://github.com/nf-core/test-datasets/raw/viralrecon/genome/kraken2/kraken2_hs22.tar.gz'
-    assemblers                  ='spades'
+    assemblers                  = 'spades'
 
     skip_metagenomic_diversity  = true
     save_databases              = true

diff --git a/modules.json b/modules.json
@@ -142,6 +142,11 @@
                         "git_sha": "3a5fef109d113b4997c9822198664ca5f2716208",
                         "installed_by": ["modules"]
                     },
+                    "humid": {
+                        "branch": "master",
+                        "git_sha": "c4174581360d2bc32c08cd39cb6a1d1c82212ff6",
+                        "installed_by": ["modules"]
+                    },
                     "ivar/consensus": {
                         "branch": "master",
                         "git_sha": "3f5420aa22e00bd030a2556dfdffc9e164ec0ec5",

diff --git a/modules/nf-core/humid/main.nf b/modules/nf-core/humid/main.nf
diff --git a/modules/nf-core/humid/meta.yml b/modules/nf-core/humid/meta.yml
diff --git a/modules/nf-core/humid/tests/main.nf.test b/modules/nf-core/humid/tests/main.nf.test
diff --git a/modules/nf-core/humid/tests/main.nf.test.snap b/modules/nf-core/humid/tests/main.nf.test.snap
diff --git a/modules/nf-core/humid/tests/nextflow.config b/modules/nf-core/humid/tests/nextflow.config
diff --git a/modules/nf-core/humid/tests/tags.yml b/modules/nf-core/humid/tests/tags.yml
diff --git a/nextflow.config b/nextflow.config
@@ -23,7 +23,8 @@ params {
     with_umi                    = false
     skip_umi_extract            = true
     umi_discard_read            = 0 // 0 = no discard
-    skip_humid                  = false
+    umi_separator               = ":"
+    umi_deduplicate             = 'read'
     humid_mismatches            = 2
 
     // > Trim
@@ -110,7 +111,6 @@ params {
     intermediate_mapper             = mapper
     iterative_refinement_cycles     = 2
     deduplicate                     = true
-    umi_separator                   = ":"
     variant_caller                  = 'ivar' // ivar, bcftools
     call_intermediate_variants      = false
     intermediate_variant_caller     = variant_caller

diff --git a/subworkflows/local/bam_deduplicate.nf b/subworkflows/local/bam_deduplicate.nf
@@ -18,7 +18,7 @@ workflow BAM_DEDUPLICATE {
     reference   = bam_ref_fai.map{meta, bam, fasta, fai -> [ meta, fasta ] }
     faidx       = bam_ref_fai.map{meta, bam, fasta, fai -> [ meta, fai ] }
 
-    if ( umi ) {
+    if ( params.with_umi && ['mapping','both'].contains(params.umi_deduplicate) ) {
             SAMTOOLS_INDEX( bam )
             ch_bam_bai  = bam.join(SAMTOOLS_INDEX.out.bai, by: [0])
             ch_versions = ch_versions.mix(SAMTOOLS_INDEX.out.versions)

diff --git a/subworkflows/local/preprocessing_illumina.nf b/subworkflows/local/preprocessing_illumina.nf
@@ -80,7 +80,7 @@ workflow PREPROCESSING_ILLUMINA {
         .set { failed_reads }
 
     // deduplicate UMI's with HUMID
-    if (params.with_umi && !params.skip_humid) {
+    if (params.with_umi && ['read', 'both'].contains(params.umi_deduplicate) && params.deduplicate ) {
         HUMID (
             ch_reads_trim,
             [[:],[]]