feat: add outbreaks.json (#484)

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

@@ -1,6 +1,9 @@
+import { MDXRemoteSerializeResult } from "next-mdx-remote";
 import { WorkflowUrlParameter } from "../../../../utils/galaxy-api/entities";
 import {
   ORGANISM_PLOIDY,
+  OUTBREAK_PRIORITY,
+  OUTBREAK_RESOURCE_TYPE,
   WORKFLOW_PARAMETER_VARIABLE,
   WORKFLOW_PLOIDY,
 } from "./schema-entities";
@@ -72,6 +75,21 @@ export interface EntitiesResponsePagination {
   total: number;
 }
 
+export interface Outbreak {
+  active: boolean;
+  description: MDXRemoteSerializeResult;
+  highlight_descendant_taxonomy_ids: number[] | null;
+  priority: OUTBREAK_PRIORITY;
+  resources: OutbreakResource[];
+  taxonomy_id: number;
+}
+
+export interface OutbreakResource {
+  title: string;
+  type: OUTBREAK_RESOURCE_TYPE;
+  url: string;
+}
+
 export interface WorkflowCategory {
   category: string;
   description: string;

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

@@ -1,5 +1,7 @@
 export {
   OrganismPloidy as ORGANISM_PLOIDY,
+  OutbreakPriority as OUTBREAK_PRIORITY,
+  OutbreakResourceType as OUTBREAK_RESOURCE_TYPE,
   WorkflowParameterVariable as WORKFLOW_PARAMETER_VARIABLE,
   WorkflowPloidy as WORKFLOW_PLOIDY,
 } from "../../../../../catalog/schema/generated/schema";

catalog/build/ts/build-assemblies.ts

@@ -0,0 +1,90 @@
+import { BRCDataCatalogGenome } from "../../../app/apis/catalog/brc-analytics-catalog/common/entities";
+import { getGenomeId } from "../../../app/apis/catalog/brc-analytics-catalog/common/utils";
+import { Organisms as SourceOrganisms } from "../../schema/generated/schema";
+import { SOURCE_GENOME_KEYS } from "./constants";
+import { SourceGenome } from "./entities";
+import {
+  defaultStringToNone,
+  parseBoolean,
+  parseList,
+  parseNumber,
+  parseNumberOrNull,
+  parseStringOrNull,
+  readValuesFile,
+  readYamlFile,
+  verifyUniqueIds,
+} from "./utils";
+
+const SOURCE_PATH_GENOMES = "catalog/build/intermediate/genomes-from-ncbi.tsv";
+const SOURCE_PATH_ORGANISMS = "catalog/source/organisms.yml";
+
+export async function buildAssemblies(): Promise<BRCDataCatalogGenome[]> {
+  const sourceRows = await readValuesFile<SourceGenome>(
+    SOURCE_PATH_GENOMES,
+    undefined,
+    SOURCE_GENOME_KEYS
+  );
+  const sourceOrganisms = await readYamlFile<SourceOrganisms>(
+    SOURCE_PATH_ORGANISMS
+  );
+  const sourceOrganismsByTaxonomyId = new Map(
+    sourceOrganisms.organisms.map((sourceOrganism) => [
+      String(sourceOrganism.taxonomy_id),
+      sourceOrganism,
+    ])
+  );
+  const mappedRows: BRCDataCatalogGenome[] = [];
+  for (const row of sourceRows) {
+    const ploidy = sourceOrganismsByTaxonomyId.get(
+      row.speciesTaxonomyId
+    )?.ploidy;
+    if (ploidy === undefined) {
+      console.log(
+        `Skipping assembly ${row.accession} [tax_id: ${row.speciesTaxonomyId}] - ploidy not found`
+      );
+      continue;
+    }
+    const taxonomicLevelStrain =
+      row.taxonomicLevelStrain ||
+      (row.strain
+        ? `${row.taxonomicLevelSpecies} strain ${row.strain}`
+        : "None");
+    mappedRows.push({
+      accession: row.accession,
+      annotationStatus: parseStringOrNull(row.annotationStatus),
+      chromosomes: parseNumberOrNull(row.chromosomeCount),
+      commonName: parseStringOrNull(row.commonName),
+      coverage: parseStringOrNull(row.coverage),
+      gcPercent: parseNumberOrNull(row.gcPercent),
+      geneModelUrl: parseStringOrNull(row.geneModelUrl),
+      isRef: parseBoolean(row.isRef),
+      length: parseNumber(row.length),
+      level: row.level,
+      lineageTaxonomyIds: parseList(row.lineageTaxonomyIds),
+      ncbiTaxonomyId: row.taxonomyId,
+      ploidy,
+      scaffoldCount: parseNumberOrNull(row.scaffoldCount),
+      scaffoldL50: parseNumberOrNull(row.scaffoldL50),
+      scaffoldN50: parseNumberOrNull(row.scaffoldN50),
+      speciesTaxonomyId: row.speciesTaxonomyId,
+      strainName: parseStringOrNull(row.strain),
+      taxonomicGroup: row.taxonomicGroup ? row.taxonomicGroup.split(",") : [],
+      taxonomicLevelClass: defaultStringToNone(row.taxonomicLevelClass),
+      taxonomicLevelDomain: defaultStringToNone(row.taxonomicLevelDomain),
+      taxonomicLevelFamily: defaultStringToNone(row.taxonomicLevelFamily),
+      taxonomicLevelGenus: defaultStringToNone(row.taxonomicLevelGenus),
+      taxonomicLevelKingdom: defaultStringToNone(row.taxonomicLevelKingdom),
+      taxonomicLevelOrder: defaultStringToNone(row.taxonomicLevelOrder),
+      taxonomicLevelPhylum: defaultStringToNone(row.taxonomicLevelPhylum),
+      taxonomicLevelRealm: defaultStringToNone(row.taxonomicLevelRealm),
+      taxonomicLevelSpecies: defaultStringToNone(row.taxonomicLevelSpecies),
+      taxonomicLevelStrain,
+      ucscBrowserUrl: parseStringOrNull(row.ucscBrowser),
+    });
+  }
+  const sortedRows = mappedRows.sort((a, b) =>
+    a.accession.localeCompare(b.accession)
+  );
+  verifyUniqueIds("assembly", sortedRows, getGenomeId);
+  return sortedRows;
+}