Simplify HMMER inputs

sanger-tol · Nov 28, 2024 · 83f586a · 83f586a
1 parent c207428
commit 83f586a
Show file tree

Hide file tree

Showing 9 changed files with 63 additions and 47 deletions.
diff --git a/.nf-core.yml b/.nf-core.yml
@@ -37,9 +37,6 @@ lint:
   multiqc_config:
     - report_comment
   nextflow_config:
-    - config_default:
-      - params.hmm_gtdb_pfam
-      - params.hmm_gtdb_tigrfam
     - process.cpus
     - process.memory
     - process.time

diff --git a/conf/modules.config b/conf/modules.config
@@ -79,16 +79,10 @@ process {
         ext.prefix = { "${meta.id}.hic" }
     }
 
-    withName: 'HMMSEARCH_GTDB_PFAM' {
+    withName: 'HMMER_HMMSEARCH' {
         ext.args   = { "" }
-        ext.prefix = { "${meta.id}_${meta.assembler}_pfam" }
-        tag        = { "${meta.id}_${meta.assembler}" }
-    }
-
-    withName: 'HMMSEARCH_GTDB_TIGRFAM' {
-        ext.args   = { "" }
-        ext.prefix = { "${meta.id}_${meta.assembler}_tigrfam" }
-        tag        = { "${meta.id}_${meta.assembler}" }
+        ext.prefix = { "${meta.id}_${meta.assembler}_${hmmfile}" }
+        tag        = { "${meta.id}_${meta.assembler}_${hmmfile}" }
     }
 
     withName: 'MAGSCOT_MAGSCOT' {
@@ -175,7 +169,7 @@ process {
     }
 
     withName: 'PYRODIGAL' {
-        ext.args   = { "" }
+        ext.args   = { "-p meta" }
         ext.prefix = { "${meta.id}_${meta.assembler}" }
         tag        = { "${meta.id}_${meta.assembler}" }
         publishDir = [

diff --git a/modules/local/magscot/magscot/main.nf b/modules/local/magscot/magscot/main.nf
@@ -26,5 +26,10 @@ process MAGSCOT_MAGSCOT {
     mv MAGScoT.scores.out ${prefix}.MAGScoT.scores.out
     mv MAGScoT.refined.out ${prefix}.MAGScoT.refined.out
     mv MAGScoT.refined.contig_to_bin.out ${prefix}.MAGScoT.refined.contig_to_bin.out
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        magscot: 1.1
+    END_VERSIONS
     """
 }
diff --git a/modules/nf-core/pyrodigal/main.nf b/modules/nf-core/pyrodigal/main.nf
diff --git a/nextflow_schema.json b/nextflow_schema.json
@@ -99,12 +99,10 @@
                     "default": true
                 },
                 "hmm_gtdb_pfam": {
-                    "type": "string",
-                    "default": "${baseDir}/assets/gtdbtk_rel207_Pfam-A.hmm"
+                    "type": "string"
                 },
                 "hmm_gtdb_tigrfam": {
-                    "type": "string",
-                    "default": "${baseDir}/assets/gtdbtk_rel207_tigrfam.hmm"
+                    "type": "string"
                 }
             }
         },

diff --git a/subworkflows/local/bin_refinement.nf b/subworkflows/local/bin_refinement.nf
@@ -2,8 +2,7 @@ include { DASTOOL_DASTOOL                           } from '../../modules/nf-cor
 include { DASTOOL_FASTATOCONTIG2BIN                 } from '../../modules/nf-core/dastool/fastatocontig2bin/main'
 include { GAWK as PROCESS_HMM_TBLOUT                } from '../../modules/nf-core/gawk/main'
 include { GAWK as MAGSCOT_PROCESS_CONTIG2BIN        } from '../../modules/nf-core/gawk/main'
-include { HMMER_HMMSEARCH as HMMSEARCH_GTDB_PFAM    } from '../../modules/nf-core/hmmer/hmmsearch/main'
-include { HMMER_HMMSEARCH as HMMSEARCH_GTDB_TIGRFAM } from '../../modules/nf-core/hmmer/hmmsearch/main'
+include { HMMER_HMMSEARCH                           } from '../../modules/nf-core/hmmer/hmmsearch/main'
 include { MAGSCOT_CONTIG2BIN2FASTA                  } from '../../modules/local/magscot/contig2bin2fasta/main'
 include { MAGSCOT_MAGSCOT                           } from '../../modules/local/magscot/magscot/main'
 include { PYRODIGAL                                 } from '../../modules/nf-core/pyrodigal/main'
@@ -18,21 +17,19 @@ workflow BIN_REFINEMENT {
     ch_refined_bins = Channel.empty()
 
     DASTOOL_FASTATOCONTIG2BIN(bins, 'fa')
+    ch_versions = ch_versions.mix(DASTOOL_FASTATOCONTIG2BIN.out.versions)
 
     if(params.enable_dastool) {
         ch_contig2bins_to_merge = DASTOOL_FASTATOCONTIG2BIN.out.fastatocontig2bin
-            | map {meta, tsv -> [meta.subMap(['id', 'assembler']), tsv] }
+            | map {meta, tsv -> [meta - meta.subMap(['binner']), tsv] }
             | groupTuple(by: 0)
 
         ch_dastool_input = assemblies
             | combine(ch_contig2bins_to_merge, by: 0)
 
         DASTOOL_DASTOOL(ch_dastool_input, [], [])
 
-        ch_versions = ch_versions.mix(
-            DASTOOL_FASTATOCONTIG2BIN.out.versions,
-            DASTOOL_DASTOOL.out.versions
-        )
+        ch_versions = ch_versions.mix(DASTOOL_DASTOOL.out.versions)
 
         ch_dastool_bins = DASTOOL_DASTOOL.out.bins
             | map {meta, fasta -> [ meta + [binner: "DASTool"], fasta ]}
@@ -43,21 +40,20 @@ workflow BIN_REFINEMENT {
     if(params.enable_magscot) {
         PYRODIGAL(assemblies, 'gff')
 
-        ch_hmmsearch_gtdb_pfam_input = PYRODIGAL.out.faa
-            | map { meta, faa ->
-                [meta, file(params.hmm_gtdb_pfam), faa, false, true, false]
-            }
+        ch_magscot_gtdb_hmm_db = Channel.of(
+            file(params.hmm_gtdb_pfam),
+            file(params.hmm_gtdb_tigrfam)
+        )
 
-        ch_hmmsearch_gtdb_tigrfam_input = PYRODIGAL.out.faa
-            | map { meta, faa ->
-                [meta, file(params.hmm_gtdb_tigrfam), faa, false, true, false]
+        ch_hmmsearch_gtdb_input = PYRODIGAL.out.faa
+            | combine(ch_magscot_gtdb_hmm_db)
+            | map { meta, faa, hmmfile ->
+                [ meta, hmmfile, faa, false, true, false ]
             }
 
-        HMMSEARCH_GTDB_PFAM(ch_hmmsearch_gtdb_pfam_input)
-        HMMSEARCH_GTDB_TIGRFAM(ch_hmmsearch_gtdb_tigrfam_input)
+        HMMER_HMMSEARCH(ch_hmmsearch_gtdb_input)
 
-        ch_hmm_output = HMMSEARCH_GTDB_PFAM.out.target_summary
-            | mix(HMMSEARCH_GTDB_TIGRFAM.out.target_summary)
+        ch_hmm_output = HMMER_HMMSEARCH.out.target_summary
             | groupTuple(by: 0)
 
         PROCESS_HMM_TBLOUT(ch_hmm_output, [])
@@ -68,7 +64,7 @@ workflow BIN_REFINEMENT {
         )
 
         ch_magscot_contig2bin = MAGSCOT_PROCESS_CONTIG2BIN.out.output
-            | map { meta, c2b -> [ meta.subMap(['id', 'assembler']), c2b ] }
+            | map { meta, c2b -> [ meta - meta.subMap(['binner']), c2b ] }
             | groupTuple(by: 0)
 
         ch_magscot_input = PROCESS_HMM_TBLOUT.out.output
@@ -85,6 +81,16 @@ workflow BIN_REFINEMENT {
             | map { meta, rbins -> [ meta + [binner: "magscot"], rbins] }
 
         ch_refined_bins = ch_refined_bins.mix(ch_magscot_bins)
+
+        ch_versions = ch_versions
+            | mix(
+                PYRODIGAL.out.versions,
+                HMMER_HMMSEARCH.out.versions,
+                PROCESS_HMM_TBLOUT.out.versions,
+                MAGSCOT_PROCESS_CONTIG2BIN.out.versions,
+                MAGSCOT_MAGSCOT.out.versions,
+                MAGSCOT_CONTIG2BIN2FASTA.out.versions
+            )
     }
 
     emit:

diff --git a/subworkflows/local/prepare_data.nf b/subworkflows/local/prepare_data.nf
@@ -26,6 +26,12 @@ workflow PREPARE_DATA {
     HIC_TO_FASTQ(ch_hic_cram, false)
     ch_hic_fastq = HIC_TO_FASTQ.out.fastq
 
+    ch_versions = ch_versions
+        | mix(
+            MERGE_HIC_CRAM.out.versions,
+            HIC_TO_FASTQ.out.versions
+        )
+
     emit:
     hic_reads = ch_hic_fastq
     versions  = ch_versions

diff --git a/subworkflows/local/read_mapping.nf b/subworkflows/local/read_mapping.nf
@@ -25,6 +25,13 @@ workflow READ_MAPPING {
 
         SORT_HIC_BAM(BWAMEM2_MEM.out.bam, [[],[]])
         ch_hic_bam = SORT_HIC_BAM.out.bam
+
+        ch_versions = ch_versions
+            | mix(
+                BWAMEM2_INDEX.out.versions,
+                BWAMEM2_MEM.out.versions,
+                SORT_HIC_BAM.out.versions
+            )
     } else {
         ch_hic_bam   = Channel.empty()
     }
@@ -47,9 +54,6 @@ workflow READ_MAPPING {
 
     ch_versions = ch_versions
         | mix(
-            BWAMEM2_INDEX.out.versions,
-            BWAMEM2_MEM.out.versions,
-            SORT_HIC_BAM.out.versions,
             MINIMAP2_ALIGN.out.versions,
             COVERM_CONTIG.out.versions
         )

diff --git a/workflows/longreadmag.nf b/workflows/longreadmag.nf
@@ -34,6 +34,7 @@ workflow LONGREADMAG {
     PREPARE_DATA(
         hic_cram
     )
+    ch_versions = ch_versions.mix(PREPARE_DATA.out.versions)
 
     if(params.enable_assembly) {
         ASSEMBLY(pacbio_fasta)
@@ -44,6 +45,7 @@ workflow LONGREADMAG {
             pacbio_fasta,
             PREPARE_DATA.out.hic_reads
         )
+        ch_versions = ch_versions.mix(READ_MAPPING.out.versions)
 
         if(params.enable_binning) {
             BINNING(
@@ -60,18 +62,21 @@ workflow LONGREADMAG {
                     BINNING.out.bins
                 )
             }
+
+            ch_bins = BINNING.out.bins
+                | BIN_REFINEMENT.out.refined_bins
         }
     }
     //
     // Collate and save software versions
     //
-    softwareVersionsToYAML(ch_versions)
-        .collectFile(
-            storeDir: "${params.outdir}/pipeline_info",
-            name:  ''  + 'pipeline_software_' +  'mqc_'  + 'versions.yml',
-            sort: true,
-            newLine: true
-        ).set { ch_collated_versions }
+    // softwareVersionsToYAML(ch_versions)
+    //     .collectFile(
+    //         storeDir: "${params.outdir}/pipeline_info",
+    //         name:  ''  + 'pipeline_software_' +  'mqc_'  + 'versions.yml',
+    //         sort: true,
+    //         newLine: true
+    //     ).set { ch_collated_versions }
 
 
     // //