initial conversion of contigsTaxonomyAssignment to subworkflow

workflows/Nextflow/main.nf

@@ -7,6 +7,7 @@ include {HOSTREMOVAL} from './modules/hostRemoval/hostRemoval.nf'
 include {ASSEMBLY} from './modules/runAssembly/runAssembly.nf'
 include {READSTOCONTIGS} from './modules/runReadsToContig/runReadsToContig.nf'
 include {READSTAXONOMYASSIGNMENT} from './modules/readsTaxonomyAssignment/readsTaxonomyAssignment.nf'
+include {CONTIGSTAXONOMYASSIGNMENT} from '.modules/contigsTaxonomyAssignment/contigsTaxonomyAssignment.nf'
 
 workflow {
 
@@ -50,8 +51,15 @@ workflow {
         READSTOCONTIGS(params.r2c.plus(params.shared), paired, unpaired, contigs)
     }
 
+    //should always run if contigs were provided or generated
+    READSTOCONTIGS(params.r2c.plus(params.shared), paired, unpaired, contigs)
+
     if(params.modules.readsTaxonomyAssignment) {
         READSTAXONOMYASSIGNMENT(params.readsTaxonomy.plus(params.shared).plus(params.faqcs.minLen), paired, unpaired, avgLen)
     }
 
+    if(params.modules.contigsTaxonomyAssignment) {
+        CONTIGSTAXONOMYASSIGNMENT(params.contigsTaxonomy.plus(params.shared), contigs, READSTOCONTIGS.out.)
+    }
+
 }

workflows/Nextflow/modules/contigsTaxonomyAssignment/contigsTaxonomyAssignment.nf

@@ -0,0 +1,89 @@
+//André Watson
+//Aug 2024
+//apwat @ lanl.gov
+
+//base process. Takes a FASTA file containing contigs and performs taxonomic analysis with MICCR (https://github.com/chienchi/miccr).
+process contigTaxonomy {
+    label 'cta'
+    publishDir(
+        path: "${settings["outDir"]}/AssemblyBasedAnalysis/Taxonomy",
+        mode: 'copy'
+    )
+    input:
+    val settings
+    path contigs
+
+    output:
+    path "${settings["projName"]}.log"
+    path "log.txt"
+    path "${settings["projName"]}.lca_ctg.tsv", emit: taxLcaResult
+    path "${settings["projName"]}.ctg.tsv", emit: taxResult
+    path "${settings["projName"]}.unclassified.fasta"
+    path "${settings["projName"]}.paf"
+
+    script:
+    """
+    miccr.py -x asm10 -d /venv/database/miccrDB/NCBI-Bacteria-Virus.fna.mmi -t ${settings["cpus"]} -p ${settings["projName"]} -i $contigs 1>log.txt 2>&1 
+    get_unclassified_fasta.pl -in $contigs -classified ${settings["projName"]}.lca_ctg.tsv -output ${settings["projName"]}.unclassified.fasta -log log.txt
+    """
+}
+
+//adds multi-level taxonomic classification to results file. Takes in a .ctg.tsv file produced by MICCR.
+process addLineage {
+    label 'cta'
+    publishDir(
+        path: "${settings["outDir"]}/AssemblyBasedAnalysis/Taxonomy",
+        mode: 'copy'
+    )
+
+    input:
+    val settings
+    path taxResult
+
+    output:
+    path "*.lineage", emit: lineage
+
+    script:
+    //add_lineage.py is from MICCR repo in container
+    """
+    add_lineage.py .venv/database/miccrDB/ $taxResult > ${taxResult.name}.lineage
+    """
+}
+
+//creates taxonomy classification graphs. Takes lineage file, .lca_ctg.tsv file produced by MICCR, 
+//and a coverage table (see https://github.com/chienchi/miccr/blob/master/utils/README.md), or from workflow runReadsToContig 
+process plotAndTable {
+    label 'cta'
+    publishDir(
+        path: "${settings["outDir"]}/AssemblyBasedAnalysis/Taxonomy",
+	mode: 'copy'
+    )
+    input:
+    val settings
+    path lineage
+    path covTable
+    path lcaResult
+
+    output:
+    path "${settings["projName"]}.ctg_class.LCA.json"
+    path "summary_by_*.txt"
+    path "*.pdf"
+
+    script:
+    """
+    classification_plot.R $lineage ${settings["projName"]} $covTable
+    tab2Json_for_dataTable.pl -project_dir $${settings["outDir"]} -mode contig -limit ${settings["rowLimit"]} $lcaResult > ${settings["projName"]}.ctg_class.LCA.json
+    """
+}
+
+workflow CONTIGSTAXONOMYASSIGNMENT {
+    take:
+    settings
+    contigs
+    coverageTable
+
+    main:
+    contigTaxonomy(settings, contigs)
+    addLineage(settings, contigTaxonomy.out.taxResult)
+    plotAndTable(settings, addLineage.out.lineage, coverageTable, contigTaxonomy.out.taxLcaResult)
+}

workflows/Nextflow/modules/runReadsToContig/runReadsToContig.nf

@@ -101,7 +101,7 @@ process makeCoverageTable {
     output:
     path "contigs_stats.txt"
     path "contigs_stats.pdf"
-    path "*_coverage.table.json"
+    path "*_coverage.table.json", emit: coverageTable
 
     script:
     def rowLimit = settings["rowLimit"] != null ? "${settings["rowLimit"]} " : "3000"
@@ -157,4 +157,9 @@ workflow READSTOCONTIGS {
         extractUnmapped(settings, validationAlignment.out.sortedBam, validationAlignment.out.logFile)
     }
 
+    covTable = makeCoverageTable.out.coverageTable
+    emit:
+    covTable
+
+
 }

workflows/Nextflow/nextflow.config

@@ -122,6 +122,10 @@ params {
 
     }
 
+    contigsTaxonomy {
+        rowLimit = 3000
+    }
+
 
 }
 
@@ -130,9 +134,11 @@ apptainer {
     enabled = true
     pullTimeout = "1 hour"
     //binds are installation-dependent binds to databases
+    //not all bound paths are accessed by all containers
     runOptions = "--compat --cleanenv --home /media/volume/sdd/nextflow \
                     --bind /media/volume/sdd/nextflow/database:/venv/bin/../../../database \
-                    --bind /media/volume/sdd/nextflow/krona_dbs:/venv/opt/krona/taxonomy"
+                    --bind /media/volume/sdd/nextflow/krona_dbs:/venv/opt/krona/taxonomy \
+                    --bind /media/volume/sdb/nextflow/database/miccrDB:/venv/database/miccrDB"
 }
 
 
@@ -157,7 +163,10 @@ process {
         container = 'apwat/run_r2c:1.6'
     }
     withLabel: 'rta' {
-        container = "apwat/reads_taxonomy:1.13"
+        container = 'apwat/reads_taxonomy:1.13'
+    }
+    withLabel: 'cta' {
+        container = 'apwat/contigs_taxonomy:1.10'
     }
 }