feat: adding organism specific workflows (#357)

app/apis/catalog/brc-analytics-catalog/common/entities.ts

@@ -4,6 +4,12 @@ import {
   WORKFLOW_PLOIDY,
 } from "./schema-entities";
 
+export interface WorkflowUrlParameter {
+  ext: string;
+  src: string;
+  url: string;
+}
+
 export type BRCCatalog = BRCDataCatalogGenome;
 
 export interface BRCDataCatalogGenome {
@@ -87,5 +93,6 @@ export interface Workflow {
 
 export interface WorkflowParameter {
   key: string;
-  variable: WORKFLOW_PARAMETER_VARIABLE;
+  url_spec?: WorkflowUrlParameter;
+  variable?: WORKFLOW_PARAMETER_VARIABLE;
 }

app/utils/galaxy-api.ts

@@ -1,5 +1,8 @@
 import { WORKFLOW_PARAMETER_VARIABLE } from "../apis/catalog/brc-analytics-catalog/common/schema-entities";
-import { WorkflowParameter } from "../apis/catalog/brc-analytics-catalog/common/entities";
+import {
+  WorkflowParameter,
+  WorkflowUrlParameter,
+} from "../apis/catalog/brc-analytics-catalog/common/entities";
 import ky from "ky";
 import { GALAXY_ENVIRONMENT } from "site-config/common/galaxy";
 
@@ -11,7 +14,7 @@ interface WorkflowLandingsBody {
 }
 
 type WorkflowLandingsBodyRequestState = {
-  [key: string]: { [key: string]: string } | string;
+  [key: string]: string | WorkflowUrlParameter;
 };
 
 interface WorkflowLanding {
@@ -106,15 +109,19 @@ function getWorkflowLandingsRequestState(
   parameters: WorkflowParameter[]
 ): WorkflowLandingsBodyRequestState {
   const result: WorkflowLandingsBodyRequestState = {};
-  parameters.forEach(({ key, variable }) => {
-    const maybeParam = paramVariableToRequestValue(
-      variable,
-      geneModelUrl,
-      referenceGenome
-    );
-    if (maybeParam !== undefined) {
-      result[key] = maybeParam;
+  for (const { key, url_spec, variable } of parameters) {
+    if (url_spec) {
+      // If url_spec is provided, use it directly
+      result[key] = url_spec;
+    } else if (variable) {
+      // Otherwise, use the variable to determine the value
+      const value = paramVariableToRequestValue(
+        variable,
+        geneModelUrl,
+        referenceGenome
+      );
+      if (value !== undefined) result[key] = value;
     }
-  });
+  }
   return result;
 }

catalog/build/py/generated_schema/schema.py

@@ -94,6 +94,7 @@ class WorkflowCategoryId(str, Enum):
     ASSEMBLY = "ASSEMBLY"
     GENOME_COMPARISONS = "GENOME_COMPARISONS"
     PROTEIN_FOLDING = "PROTEIN_FOLDING"
+    CONSENSUS_SEQUENCES = "CONSENSUS_SEQUENCES"
     OTHER = "OTHER"
 
 
@@ -211,9 +212,21 @@ class WorkflowParameter(ConfiguredBaseModel):
 
     key: str = Field(default=..., description="""The key in which the parameter will be set.""", json_schema_extra = { "linkml_meta": {'alias': 'key', 'domain_of': ['WorkflowParameter']} })
     variable: Optional[WorkflowParameterVariable] = Field(default=None, description="""A variable to substitute in as the value of the parameter.""", json_schema_extra = { "linkml_meta": {'alias': 'variable', 'domain_of': ['WorkflowParameter']} })
+    url_spec: Optional[WorkflowUrlSpec] = Field(default=None, description="""A direct URL specification for the parameter.""", json_schema_extra = { "linkml_meta": {'alias': 'url_spec', 'domain_of': ['WorkflowParameter']} })
     type_guide: Optional[Any] = Field(default=None, description="""Arbitrary data describing the type of the parameter, intended only as convenient reference for maintainers.""", json_schema_extra = { "linkml_meta": {'alias': 'type_guide', 'domain_of': ['WorkflowParameter']} })
 
 
+class WorkflowUrlSpec(ConfiguredBaseModel):
+    """
+    A URL specification for a workflow parameter.
+    """
+    linkml_meta: ClassVar[LinkMLMeta] = LinkMLMeta({'from_schema': 'https://github.com/galaxyproject/brc-analytics/blob/main/catalog/schema/workflows.yaml#'})
+
+    ext: str = Field(default=..., description="""The file extension of the URL.""", json_schema_extra = { "linkml_meta": {'alias': 'ext', 'domain_of': ['WorkflowUrlSpec']} })
+    src: str = Field(default=..., description="""The source type, typically 'url'.""", json_schema_extra = { "linkml_meta": {'alias': 'src', 'domain_of': ['WorkflowUrlSpec']} })
+    url: str = Field(default=..., description="""The URL to the resource.""", json_schema_extra = { "linkml_meta": {'alias': 'url', 'domain_of': ['WorkflowUrlSpec']} })
+
+
 # Model rebuild
 # see https://pydantic-docs.helpmanual.io/usage/models/#rebuilding-a-model
 Assemblies.model_rebuild()
@@ -225,4 +238,5 @@ class WorkflowParameter(ConfiguredBaseModel):
 Workflows.model_rebuild()
 Workflow.model_rebuild()
 WorkflowParameter.model_rebuild()
+WorkflowUrlSpec.model_rebuild()
 

catalog/build/py/iwc_manifest_to_workflows_yaml.py

@@ -16,6 +16,7 @@
     "Transcriptomics": WorkflowCategoryId.TRANSCRIPTOMICS,
     "Epigenetics": WorkflowCategoryId.REGULATION,
     "Genome assembly": WorkflowCategoryId.ASSEMBLY,
+    "Virology": WorkflowCategoryId.CONSENSUS_SEQUENCES
 }
 MANIFEST_SOURCE_OF_TRUTH = ("trs_id", "workflow_name", "categories", "workflow_description")
 

catalog/build/ts/build-catalog.ts

@@ -198,8 +198,10 @@ function buildWorkflow(
   }: SourceWorkflow
 ): void {
   const parameters = [];
-  for (const { key, variable } of sourceParameters) {
+  for (const { key, url_spec, variable } of sourceParameters) {
+    // Add parameter if either variable or url_spec is defined
     if (variable) parameters.push({ key, variable });
+    else if (url_spec) parameters.push({ key, url_spec });
   }
   const workflow: Workflow = {
     parameters,

catalog/output/workflows.json

@@ -20,6 +20,61 @@
         "trsId": "#workflow/github.com/iwc-workflows/haploid-variant-calling-wgs-pe/main/versions/v0.1",
         "workflowDescription": "Workflow for variant analysis against a reference genome in GenBank format",
         "workflowName": "Paired end variant calling in haploid system"
+      },
+      {
+        "parameters": [
+          {
+            "key": "NC_045512.2 FASTA sequence of SARS-CoV-2",
+            "variable": "ASSEMBLY_FASTA_URL"
+          },
+          {
+            "key": "ARTIC primer BED",
+            "url_spec": {
+              "ext": "bed",
+              "src": "url",
+              "url": "https://zenodo.org/record/4555735/files/ARTIC_nCoV-2019_v3.bed"
+            }
+          },
+          {
+            "key": "ARTIC primers to amplicon assignments",
+            "url_spec": {
+              "ext": "tabular",
+              "src": "url",
+              "url": "https://zenodo.org/record/4555735/files/ARTIC_amplicon_info_v3.tsv"
+            }
+          }
+        ],
+        "ploidy": "HAPLOID",
+        "taxonomyId": "694009",
+        "trsId": "#workflow/github.com/iwc-workflows/sars-cov-2-pe-illumina-artic-variant-calling/COVID-19-PE-ARTIC-ILLUMINA/versions/v0.5.4",
+        "workflowDescription": "The workflow for Illumina-sequenced ARTIC data builds on the RNASeq workflow for paired-end data using the same steps for mapping and variant calling, but adds extra logic for trimming ARTIC primer sequences off reads with the ivar package. In addition, this workflow uses ivar also to identify amplicons affected by ARTIC primer-binding site mutations and tries to exclude reads derived from such tainted amplicons when calculating allele-frequencies of other variants.",
+        "workflowName": "COVID-19: variation analysis on ARTIC PE data"
+      },
+      {
+        "parameters": [
+          {
+            "key": "NC_045512.2 FASTA sequence of SARS-CoV-2",
+            "variable": "ASSEMBLY_FASTA_URL"
+          }
+        ],
+        "ploidy": "HAPLOID",
+        "taxonomyId": "694009",
+        "trsId": "#workflow/github.com/iwc-workflows/sars-cov-2-pe-illumina-wgs-variant-calling/COVID-19-PE-WGS-ILLUMINA/versions/v0.2.4",
+        "workflowDescription": "This workflows performs paired end read mapping with bwa-mem followed by sensitive variant calling across a wide range of AFs with lofreq",
+        "workflowName": "COVID-19: variation analysis on WGS PE data"
+      },
+      {
+        "parameters": [
+          {
+            "key": "NC_045512.2 FASTA sequence of SARS-CoV-2",
+            "variable": "ASSEMBLY_FASTA_URL"
+          }
+        ],
+        "ploidy": "HAPLOID",
+        "taxonomyId": "694009",
+        "trsId": "#workflow/github.com/iwc-workflows/sars-cov-2-se-illumina-wgs-variant-calling/COVID-19-SE-WGS-ILLUMINA/versions/v0.1.6",
+        "workflowDescription": "This workflows performs single end read mapping with bowtie2 followed by sensitive variant calling across a wide range of AFs with lofreq",
+        "workflowName": "COVID-19: variation analysis on WGS SE data"
       }
     ]
   },
@@ -165,6 +220,26 @@
       }
     ]
   },
+  {
+    "category": "CONSENSUS_SEQUENCES",
+    "description": "Build consensus sequences for related isolates.",
+    "name": "Consensus sequences",
+    "workflows": [
+      {
+        "parameters": [
+          {
+            "key": "Reference FASTA",
+            "variable": "ASSEMBLY_FASTA_URL"
+          }
+        ],
+        "ploidy": "HAPLOID",
+        "taxonomyId": "10244",
+        "trsId": "#workflow/github.com/iwc-workflows/pox-virus-amplicon/main/versions/v0.3",
+        "workflowDescription": "A workflow for the analysis of pox virus genomes sequenced as half-genomes (for ITR resolution) in a tiled-amplicon approach",
+        "workflowName": "Pox Virus Illumina Amplicon Workflow from half-genomes"
+      }
+    ]
+  },
   {
     "category": "ASSEMBLY",
     "description": "Assemble prokaryotic and eukaryotic genomes sequenced with a variety of technologies.",

catalog/schema/enums/workflow_category_id.yaml

@@ -12,4 +12,5 @@ enums:
       ASSEMBLY:
       GENOME_COMPARISONS:
       PROTEIN_FOLDING:
+      CONSENSUS_SEQUENCES:
       OTHER:

catalog/schema/generated/schema.ts

@@ -18,6 +18,7 @@ export enum WorkflowCategoryId {
     ASSEMBLY = "ASSEMBLY",
     GENOME_COMPARISONS = "GENOME_COMPARISONS",
     PROTEIN_FOLDING = "PROTEIN_FOLDING",
+    CONSENSUS_SEQUENCES = "CONSENSUS_SEQUENCES",
     OTHER = "OTHER",
 };
 /**
@@ -148,9 +149,24 @@ export interface WorkflowParameter {
     key: string,
     /** A variable to substitute in as the value of the parameter. */
     variable?: WorkflowParameterVariable | null,
+    /** A direct URL specification for the parameter. */
+    url_spec?: WorkflowUrlSpec | null,
     /** Arbitrary data describing the type of the parameter, intended only as convenient reference for maintainers. */
     type_guide?: Any | null,
 }
 
 
+/**
+ * A URL specification for a workflow parameter.
+ */
+export interface WorkflowUrlSpec {
+    /** The file extension of the URL. */
+    ext: string,
+    /** The source type, typically 'url'. */
+    src: string,
+    /** The URL to the resource. */
+    url: string,
+}
+
+
 

catalog/schema/generated/workflow_categories.json

@@ -52,6 +52,7 @@
                 "ASSEMBLY",
                 "GENOME_COMPARISONS",
                 "PROTEIN_FOLDING",
+                "CONSENSUS_SEQUENCES",
                 "OTHER"
             ],
             "title": "WorkflowCategoryId",

catalog/schema/generated/workflows.json

@@ -79,6 +79,7 @@
                 "ASSEMBLY",
                 "GENOME_COMPARISONS",
                 "PROTEIN_FOLDING",
+                "CONSENSUS_SEQUENCES",
                 "OTHER"
             ],
             "title": "WorkflowCategoryId",
@@ -103,6 +104,17 @@
                     ],
                     "description": "Arbitrary data describing the type of the parameter, intended only as convenient reference for maintainers."
                 },
+                "url_spec": {
+                    "anyOf": [
+                        {
+                            "$ref": "#/$defs/WorkflowUrlSpec"
+                        },
+                        {
+                            "type": "null"
+                        }
+                    ],
+                    "description": "A direct URL specification for the parameter."
+                },
                 "variable": {
                     "$ref": "#/$defs/WorkflowParameterVariable",
                     "description": "A variable to substitute in as the value of the parameter."
@@ -135,6 +147,31 @@
             "title": "WorkflowPloidy",
             "type": "string"
         },
+        "WorkflowUrlSpec": {
+            "additionalProperties": false,
+            "description": "A URL specification for a workflow parameter.",
+            "properties": {
+                "ext": {
+                    "description": "The file extension of the URL.",
+                    "type": "string"
+                },
+                "src": {
+                    "description": "The source type, typically 'url'.",
+                    "type": "string"
+                },
+                "url": {
+                    "description": "The URL to the resource.",
+                    "type": "string"
+                }
+            },
+            "required": [
+                "ext",
+                "src",
+                "url"
+            ],
+            "title": "WorkflowUrlSpec",
+            "type": "object"
+        },
         "Workflows": {
             "additionalProperties": false,
             "description": "Object containing list of workflows.",

catalog/schema/workflows.yaml

@@ -75,7 +75,27 @@ classes:
         description: A variable to substitute in as the value of the parameter.
         required: false
         range: WorkflowParameterVariable
+      url_spec:
+        description: A direct URL specification for the parameter.
+        required: false
+        range: WorkflowUrlSpec
       type_guide:
         description: Arbitrary data describing the type of the parameter, intended only as convenient reference for maintainers.
         required: false
         range: Any
+
+  WorkflowUrlSpec:
+    description: A URL specification for a workflow parameter.
+    attributes:
+      ext:
+        description: The file extension of the URL.
+        required: true
+        range: string
+      src:
+        description: The source type, typically 'url'.
+        required: true
+        range: string
+      url:
+        description: The URL to the resource.
+        required: true
+        range: string

catalog/source/workflow_categories.yml

@@ -11,6 +11,10 @@ workflow_categories:
     name: "Regulation"
     description: "Workflows for the analysis of ChIP-seq, ATAC-Seq, and beyond."
 
+  - category: "CONSENSUS_SEQUENCES"
+    name: "Consensus sequences"
+    description: "Build consensus sequences for related isolates."
+
   - category: "ASSEMBLY"
     name: "Assembly"
     description: "Assemble prokaryotic and eukaryotic genomes sequenced with a variety of technologies."