Adds Vibecheck to the Vibrio spp. subworkflow of TheiaProk

tasks/species_typing/vibrio/task_vibecheck_vibrio.wdl

@@ -0,0 +1,59 @@
+version 1.0
+
+task vibecheck_vibrio {
+  meta {
+    description: "Rapidly assigns O1 Vibrio cholerae sequencing data to transmission lineages using variant frequency demixing."
+  }
+  input {
+    File read1
+    File? read2
+    File? lineage_barcodes
+    Float? subsampling_fraction
+    Boolean skip_subsampling = false
+    String docker = "watronfire/vibecheck:2025.02.24"
+    Int disk_size = 16
+    Int cpu = 2
+    Int memory = 3
+  }
+  command <<<
+    set -ex
+
+    # Capture Vibecheck version information.
+    vibecheck -v | tee VERSION
+
+    # Unclear if this will work with single-reads, will explore and update.
+    vibecheck "~{read1}" "~{read2}"  \
+        --outdir . \
+        ~{"--barcodes " + lineage_barcodes} \
+        ~{"--subsampling_fraction " + subsampling_fraction} \
+        ~{true='--no-detect' false='' skip_subsampling}
+
+    # Parse Vibecheck results to capture lineage estimate, confidence, and any classification notes produced by Freyja.
+    python3 <<CODE
+    import csv
+    with open( "lineage_report.csv", "rt" ) as csv_file:
+      reader = csv.DictReader( csv_file )
+      line = next( reader )
+      for key in ["lineage", "confidence", "classification_notes"]:
+        with open( key.upper(), "wt" ) as outf:
+          outf.write(line[key])
+    CODE
+  >>>
+  output {
+    File vibecheck_lineage_report = "lineage_report.csv"
+    String vibecheck_top_lineage = read_string("LINEAGE")
+    Float vibecheck_confidence = read_float("CONFIDENCE")
+    String vibecheck_classification_notes = read_string("CLASSIFICATION_NOTES")
+    String vibecheck_version = read_string("VERSION")
+    String vibecheck_docker = docker
+  }
+  runtime {
+    docker: docker
+    memory: memory + " GB"
+    cpu: cpu
+    disks: "local-disk " + disk_size + " SSD"
+    disk: disk_size + " GB"
+    maxRetries: 3
+    preemptible: 0
+  }
+}

tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml

@@ -512,9 +512,9 @@
     - path: miniwdl_run/wdl/tasks/taxon_id/contamination/task_midas.wdl
       md5sum: 64caaaff5910ac0036e2659434500962
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
-      md5sum: cea2e72701662adca0b30fd48baca3ab
+      md5sum: 8d2c15e11472f0ca1df84ddcbb979cf2
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
-      md5sum: e797bbc7ceb4f62728f7ef315d9a4e40
+      md5sum: 7e66aa483f679261d1e5fd11f22e4f5d
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_pe.wdl
       contains: ["version", "QC", "output"]
     - path: miniwdl_run/workflow.log

tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml

@@ -485,7 +485,7 @@
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
       md5sum: ff0f911bbaf3876ecb139bf6c564f1df
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
-      md5sum: e797bbc7ceb4f62728f7ef315d9a4e40
+      md5sum: 7e66aa483f679261d1e5fd11f22e4f5d
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_se.wdl
       md5sum: 09d9f68b9ca8bf94b6145ff9bed2edd1
     - path: miniwdl_run/workflow.log

workflows/theiaprok/wf_theiaprok_illumina_pe.wdl

@@ -684,6 +684,12 @@ workflow theiaprok_illumina_pe {
                 "ts_mlst_pubmlst_scheme": ts_mlst.ts_mlst_pubmlst_scheme,
                 "ts_mlst_results": ts_mlst.ts_mlst_results,
                 "ts_mlst_version": ts_mlst.ts_mlst_version,
+                "vibecheck_lineage_report": merlin_magic.vibecheck_lineage_report,
+                "vibecheck_top_lineage": merlin_magic.vibecheck_top_lineage,
+                "vibecheck_confidence": merlin_magic.vibecheck_confidence,
+                "vibecheck_classification_notes": merlin_magic.vibecheck_classification_notes,
+                "vibecheck_version": merlin_magic.vibecheck_version,
+                "vibecheck_docker": merlin_magic.vibecheck_docker,
                 "virulencefinder_docker": merlin_magic.virulencefinder_docker,
                 "virulencefinder_hits": merlin_magic.virulencefinder_hits,
                 "virulencefinder_report_tsv": merlin_magic.virulencefinder_report_tsv,
@@ -1129,7 +1135,13 @@ workflow theiaprok_illumina_pe {
     String? abricate_vibrio_toxR = merlin_magic.abricate_vibrio_toxR
     String? abricate_vibrio_biotype = merlin_magic.abricate_vibrio_biotype
     String? abricate_vibrio_serogroup = merlin_magic.abricate_vibrio_serogroup
+    File? vibecheck_lineage_report = merlin_magic.vibecheck_lineage_report
+    String? vibecheck_top_lineage = merlin_magic.vibecheck_top_lineage
+    Float? vibecheck_confidence = merlin_magic.vibecheck_confidence
+    String? vibecheck_classification_notes = merlin_magic.vibecheck_classification_notes
+    String? vibecheck_version = merlin_magic.vibecheck_version
+    String? vibecheck_docker = merlin_magic.vibecheck_docker
     # export taxon table output
     String? taxon_table_status = export_taxon_table.status
   }
-}
+}

workflows/utilities/wf_merlin_magic.wdl

@@ -33,6 +33,7 @@ import "../../tasks/species_typing/streptococcus/task_poppunk_streppneumo.wdl" a
 import "../../tasks/species_typing/streptococcus/task_seroba.wdl" as seroba
 import "../../tasks/species_typing/vibrio/task_srst2_vibrio.wdl" as srst2_vibrio_task
 import "../../tasks/species_typing/vibrio/task_abricate_vibrio.wdl" as abricate_vibrio_task
+import "../../tasks/species_typing/vibrio/task_vibecheck_vibrio.wdl" as vibecheck_vibrio_task
 
 # theiaeuk
 import "../../tasks/gene_typing/variant_detection/task_snippy_gene_query.wdl" as snippy_gene_query
@@ -92,6 +93,7 @@ workflow merlin_magic {
     String? staphopia_sccmec_docker_image
     String? tbprofiler_docker_image
     String? tbp_parser_docker_image
+    String? vibecheck_docker_image
     String? virulencefinder_docker_image
     # abricate abaum options
     Int abricate_abaum_min_percent_identity = 95 # strict threshold of 95% identity for typing purposes
@@ -228,6 +230,10 @@ workflow merlin_magic {
     Float? tbp_parser_rpob449_frequency
     Float? tbp_parser_etha237_frequency
     File? tbp_parser_expert_rule_regions_bed
+    # Vibecheck options
+    File? vibecheck_lineage_barcodes
+    Float? vibecheck_subsampling_fraction
+    Boolean? vibecheck_skip_subsampling
     # virulencefinder options
     Float? virulencefinder_min_coverage
     Float? virulencefinder_min_percent_identity
@@ -626,6 +632,15 @@ workflow merlin_magic {
           gene_max_mismatch = srst2_gene_max_mismatch,
           docker = srst2_docker_image
       }
+      call vibecheck_vibrio_task.vibecheck_vibrio {
+        input:
+          read1 = select_first([read1]),
+          read2 = read2,
+          lineage_barcodes = vibecheck_lineage_barcodes,
+          subsampling_fraction = vibecheck_subsampling_fraction,
+          skip_subsampling = vibecheck_skip_subsampling,
+          docker = vibecheck_docker_image
+      }
     }
     call abricate_vibrio_task.abricate_vibrio {
       input:
@@ -1010,6 +1025,12 @@ workflow merlin_magic {
     String? abricate_vibrio_toxR = abricate_vibrio.abricate_vibrio_toxR
     String? abricate_vibrio_biotype = abricate_vibrio.abricate_vibrio_biotype
     String? abricate_vibrio_serogroup = abricate_vibrio.abricate_vibrio_serogroup
+    File? vibecheck_lineage_report = vibecheck_vibrio.vibecheck_lineage_report
+    String? vibecheck_top_lineage = vibecheck_vibrio.vibecheck_top_lineage
+    Float? vibecheck_confidence = vibecheck_vibrio.vibecheck_confidence
+    String? vibecheck_classification_notes = vibecheck_vibrio.vibecheck_classification_notes
+    String? vibecheck_version = vibecheck_vibrio.vibecheck_version
+    String? vibecheck_docker = vibecheck_vibrio.vibecheck_docker
 
     # theiaeuk
     # c auris 
@@ -1034,4 +1055,4 @@ workflow merlin_magic {
     String snippy_variants_num_variants = select_first([snippy_cauris.snippy_variants_num_variants, snippy_afumigatus.snippy_variants_num_variants, snippy_crypto.snippy_variants_num_variants, "No matching taxon detected"])
     String snippy_variants_percent_ref_coverage = select_first([snippy_cauris.snippy_variants_percent_ref_coverage, snippy_afumigatus.snippy_variants_percent_ref_coverage, snippy_crypto.snippy_variants_percent_ref_coverage, "No matching taxon detected"])
   }
-}
+}