feat: [lungmap] add lungmap projects to google datasets catalog (#4808)

__tests__/utils/schemaOrg/anvilDataset.test.ts

@@ -98,7 +98,7 @@ describe("buildAnvilDatasetJsonLd", () => {
     response.datasets[0].description = "Short.";
     const result = buildAnvilDatasetJsonLd(response, BROWSER_URL);
     expect(result!.description).toBe(
-      "Rare disease dataset — Short. — A genomic dataset in the AnVIL Data Explorer catalog."
+      "Rare disease dataset — Short. — A dataset in the AnVIL Data Explorer for NHGRI's Analysis Visualization and Informatics Lab-space."
     );
     expect(result!.description.length).toBeGreaterThanOrEqual(
       DESCRIPTION_LENGTH.MIN
@@ -110,7 +110,7 @@ describe("buildAnvilDatasetJsonLd", () => {
     response.datasets[0].description = undefined;
     const result = buildAnvilDatasetJsonLd(response, BROWSER_URL);
     expect(result!.description).toBe(
-      "Rare disease dataset — A genomic dataset in the AnVIL Data Explorer catalog."
+      "Rare disease dataset — A dataset in the AnVIL Data Explorer for NHGRI's Analysis Visualization and Informatics Lab-space."
     );
     expect(result!.description.length).toBeGreaterThanOrEqual(
       DESCRIPTION_LENGTH.MIN

__tests__/utils/schemaOrg/hcaProjectDataset.test.ts

@@ -68,7 +68,7 @@ describe("buildHcaProjectJsonLd", () => {
     expect(result!.isAccessibleForFree).toBe(true);
     expect(result!.includedInDataCatalog).toEqual({
       "@type": "DataCatalog",
-      name: "Human Cell Atlas Data Coordination Platform",
+      name: "Human Cell Atlas Data Explorer",
       url: BROWSER_URL,
     });
   });
@@ -95,7 +95,7 @@ describe("buildHcaProjectJsonLd", () => {
     response.projects[0].projectDescription = "Short.";
     const result = buildHcaProjectJsonLd(response, BROWSER_URL);
     expect(result!.description).toBe(
-      "Cells of the body — Short. — Human Cell Atlas Data Coordination Platform project."
+      "Cells of the body — Short. — A project in the Human Cell Atlas Data Explorer."
     );
     expect(result!.description.length).toBeGreaterThanOrEqual(
       DESCRIPTION_LENGTH.MIN
@@ -107,7 +107,7 @@ describe("buildHcaProjectJsonLd", () => {
     response.projects[0].projectDescription = "";
     const result = buildHcaProjectJsonLd(response, BROWSER_URL);
     expect(result!.description).toBe(
-      "Cells of the body — Human Cell Atlas Data Coordination Platform project."
+      "Cells of the body — A project in the Human Cell Atlas Data Explorer."
     );
     expect(result!.description.length).toBeGreaterThanOrEqual(
       DESCRIPTION_LENGTH.MIN

__tests__/utils/schemaOrg/lungmapProjectDataset.test.ts

@@ -0,0 +1,74 @@
+import type { ProjectsResponse } from "../../../app/apis/azul/hca-dcp/common/responses";
+import { buildLungmapProjectJsonLd } from "../../../app/utils/schemaOrg/lungmapProjectDataset";
+
+const BROWSER_URL = "https://data-browser.lungmap.net";
+
+/**
+ * Builds a minimal valid project response for the LungMAP wrapper. The full
+ * mapping is covered by `hcaProjectDataset.test.ts` (same shared core); this
+ * file only verifies the LungMAP-specific catalog identity surfaces correctly.
+ * @returns A `ProjectsResponse` shape sufficient for catalog-identity checks.
+ */
+function makeProjectsResponse(): ProjectsResponse {
+  return {
+    dates: [],
+    donorOrganisms: [],
+    entryId: "abc",
+    fileTypeSummaries: [],
+    projects: [
+      {
+        accessible: true,
+        accessions: [],
+        bionetworkName: [],
+        contributedAnalyses: {},
+        contributors: [],
+        dataUseRestriction: null,
+        duosId: null,
+        estimatedCellCount: null,
+        laboratory: [],
+        matrices: {},
+        projectDescription:
+          "A study of lung development and disease across many donors.",
+        projectId: "uuid-1",
+        projectShortname: "Lung Study",
+        projectTitle: "Lung development atlas",
+      },
+    ],
+    protocols: [],
+    samples: [],
+    specimens: [],
+    status: 200,
+  } as unknown as ProjectsResponse;
+}
+
+describe("buildLungmapProjectJsonLd", () => {
+  it("returns undefined when no project is present", () => {
+    const response = { ...makeProjectsResponse(), projects: [] };
+    expect(
+      buildLungmapProjectJsonLd(response as ProjectsResponse, BROWSER_URL)
+    ).toBeUndefined();
+  });
+
+  it("surfaces LungMAP as the catalog identity and uses the projects URL pattern", () => {
+    const result = buildLungmapProjectJsonLd(
+      makeProjectsResponse(),
+      BROWSER_URL
+    );
+    expect(result).toBeDefined();
+    expect(result!.includedInDataCatalog).toEqual({
+      "@type": "DataCatalog",
+      name: "LungMAP Data Explorer",
+      url: BROWSER_URL,
+    });
+    expect(result!.url).toBe(`${BROWSER_URL}/projects/uuid-1`);
+  });
+
+  it("pads short descriptions with the LungMAP catalog suffix", () => {
+    const response = makeProjectsResponse();
+    response.projects[0].projectDescription = "Short.";
+    const result = buildLungmapProjectJsonLd(response, BROWSER_URL);
+    expect(result!.description).toBe(
+      "Lung development atlas — Short. — A project in the LungMAP Data Explorer."
+    );
+  });
+});

app/utils/schemaOrg/anvilDataset.ts

@@ -4,7 +4,7 @@ import type { SchemaDataset } from "./types";
 import { buildDescription, uniqueNonEmpty } from "./utils";
 
 const CATALOG_NAME = "AnVIL Data Explorer";
-const DESCRIPTION_FALLBACK_SUFFIX = `A genomic dataset in the ${CATALOG_NAME} catalog.`;
+const DESCRIPTION_FALLBACK_SUFFIX = `A dataset in the AnVIL Data Explorer for NHGRI's Analysis Visualization and Informatics Lab-space.`;
 
 /**
  * Builds a Schema.org Dataset JSON-LD object for an AnVIL CMG dataset.

app/utils/schemaOrg/hcaProjectDataset.ts

@@ -1,81 +1,17 @@
-import type {
-  AccessionResponse,
-  ContributorResponse,
-  PublicationResponse,
-} from "../../apis/azul/hca-dcp/common/entities";
 import type { ProjectsResponse } from "../../apis/azul/hca-dcp/common/responses";
-import { transformAccessionURL } from "../../viewModelBuilders/azul/hca-dcp/common/accessionMapper/accessionMapper";
-import { ACCESSION_CONFIGS_BY_RESPONSE_KEY } from "../../viewModelBuilders/azul/hca-dcp/common/accessionMapper/constants";
-import { MAX_KEYWORDS } from "./constants";
-import type {
-  SchemaDataset,
-  SchemaOrganization,
-  SchemaPerson,
-  SchemaScholarlyArticle,
-} from "./types";
-import { buildDescription, uniqueNonEmpty } from "./utils";
+import type { ProjectCatalogOptions } from "./projectDataset";
+import { buildProjectJsonLd } from "./projectDataset";
+import type { SchemaDataset } from "./types";
 
-const CATALOG_NAME = "Human Cell Atlas Data Coordination Platform";
-const DESCRIPTION_FALLBACK_SUFFIX = `${CATALOG_NAME} project.`;
+const CATALOG_NAME = "Human Cell Atlas Data Explorer";
 
-/**
- * Builds the citation array from project publications. Skips entries without a
- * title. Prefers DOI for `sameAs`, falling back to the publication URL.
- * @param publications - HCA project publications.
- * @returns Array of schema.org ScholarlyArticle objects.
- */
-function buildCitations(
-  publications: PublicationResponse[]
-): SchemaScholarlyArticle[] {
-  const citations: SchemaScholarlyArticle[] = [];
-  for (const publication of publications ?? []) {
-    if (!publication.publicationTitle) continue;
-    const article: SchemaScholarlyArticle = {
-      "@type": "ScholarlyArticle",
-      headline: publication.publicationTitle,
-      name: publication.publicationTitle,
-    };
-    if (publication.doi) {
-      article.sameAs = `https://doi.org/${publication.doi}`;
-    } else if (publication.publicationUrl) {
-      article.sameAs = publication.publicationUrl;
-    }
-    citations.push(article);
-  }
-  return citations;
-}
-
-/**
- * Builds the creator array from project contributors. Skips entries without a
- * name. When the contributor has an institution, attaches it as an affiliation.
- * @param contributors - HCA project contributors.
- * @returns Array of schema.org Person objects.
- */
-function buildCreators(contributors: ContributorResponse[]): SchemaPerson[] {
-  const creators: SchemaPerson[] = [];
-  for (const contributor of contributors ?? []) {
-    if (!contributor.contactName) continue;
-    const person: SchemaPerson = {
-      "@type": "Person",
-      name: normaliseContactName(contributor.contactName),
-    };
-    if (contributor.institution) {
-      const affiliation: SchemaOrganization = {
-        "@type": "Organization",
-        name: contributor.institution,
-      };
-      person.affiliation = affiliation;
-    }
-    creators.push(person);
-  }
-  return creators;
-}
+const OPTIONS: ProjectCatalogOptions = {
+  catalogName: CATALOG_NAME,
+  descriptionFallbackSuffix: `A project in the Human Cell Atlas Data Explorer.`,
+};
 
 /**
  * Builds a Schema.org Dataset JSON-LD object for an HCA DCP project.
- *
- * Returns `undefined` when the response does not carry a project we can
- * describe (i.e. no project entity), so the caller can skip rendering.
  * @param data - HCA DCP project detail response from Azul.
  * @param browserURL - Site base URL used for canonical and catalog URLs.
  * @returns Schema.org Dataset JSON-LD object, or `undefined` if not buildable.
@@ -84,125 +20,5 @@ export function buildHcaProjectJsonLd(
   data: ProjectsResponse,
   browserURL: string
 ): SchemaDataset | undefined {
-  const project = data.projects?.[0];
-  if (!project) return undefined;
-
-  const name = project.projectTitle || project.projectShortname;
-  const description = buildDescription(
-    project.projectDescription,
-    name,
-    DESCRIPTION_FALLBACK_SUFFIX
-  );
-  const identifier = uniqueNonEmpty([
-    project.projectId,
-    ...project.accessions.flatMap((accession) =>
-      splitAccessionIds(accession.accession)
-    ),
-  ]);
-
-  const jsonLd: SchemaDataset = {
-    "@context": "https://schema.org",
-    "@type": "Dataset",
-    description,
-    identifier,
-    includedInDataCatalog: {
-      "@type": "DataCatalog",
-      name: CATALOG_NAME,
-      url: browserURL,
-    },
-    isAccessibleForFree: true,
-    name,
-    url: `${browserURL}/projects/${project.projectId}`,
-  };
-
-  const sameAs = buildSameAs(project.accessions);
-  if (sameAs.length > 0) jsonLd.sameAs = sameAs;
-
-  const keywords = buildKeywords(data);
-  if (keywords.length > 0) jsonLd.keywords = keywords;
-
-  const creator = buildCreators(project.contributors);
-  if (creator.length > 0) jsonLd.creator = creator;
-
-  const citation = buildCitations(project.publications);
-  if (citation.length > 0) jsonLd.citation = citation;
-
-  return jsonLd;
-}
-
-/**
- * Builds a keywords array by unioning biologically-meaningful fields from the
- * project's aggregated donor/sample/specimen/protocol responses.
- * @param data - HCA project detail response.
- * @returns Deduplicated keywords array.
- */
-function buildKeywords(data: ProjectsResponse): string[] {
-  const values: (string | null | undefined)[] = [];
-  for (const donor of data.donorOrganisms ?? []) {
-    values.push(...(donor.genusSpecies ?? []));
-    values.push(...(donor.disease ?? []));
-  }
-  for (const sample of data.samples ?? []) {
-    values.push(...(sample.organ ?? []));
-    values.push(...(sample.organPart ?? []));
-    values.push(...(sample.disease ?? []));
-    values.push(...(sample.sampleEntityType ?? []));
-  }
-  for (const specimen of data.specimens ?? []) {
-    values.push(...(specimen.organ ?? []));
-    values.push(...(specimen.organPart ?? []));
-    values.push(...(specimen.disease ?? []));
-  }
-  for (const protocol of data.protocols ?? []) {
-    values.push(...(protocol.libraryConstructionApproach ?? []));
-    values.push(...(protocol.instrumentManufacturerModel ?? []));
-  }
-  return uniqueNonEmpty(values).slice(0, MAX_KEYWORDS);
-}
-
-/**
- * Builds the sameAs array of external accession URLs via identifiers.org.
- * Only includes accessions whose namespace maps to a known identifier prefix.
- * @param accessions - Project accessions from the Azul response.
- * @returns Array of canonical accession URLs.
- */
-function buildSameAs(accessions: AccessionResponse[]): string[] {
-  const urls: string[] = [];
-  for (const { accession, namespace } of accessions) {
-    const prefix =
-      ACCESSION_CONFIGS_BY_RESPONSE_KEY.get(namespace)?.identifierOrgPrefix;
-    if (!prefix) continue;
-    for (const id of splitAccessionIds(accession)) {
-      const url = transformAccessionURL(id, prefix);
-      if (url) urls.push(url);
-    }
-  }
-  return uniqueNonEmpty(urls);
-}
-
-/**
- * Normalises an HCA contributor's contactName from "Last,First,Middle" to
- * "First Middle Last" for use as a Schema.org Person.name value.
- * @param contactName - Raw contactName from the Azul response.
- * @returns Human-readable contributor name.
- */
-function normaliseContactName(contactName: string): string {
-  const parts = contactName.split(",").map((part) => part.trim());
-  if (parts.length < 2) return contactName;
-  const [last, ...rest] = parts;
-  return [...rest, last].filter(Boolean).join(" ");
-}
-
-/**
- * Splits an Azul accession string into individual accession IDs. Azul returns
- * accessions as a semicolon-separated string when a project carries multiple
- * IDs under the same namespace (mirrors the split done by `mapAccessions`).
- * @param accession - Raw accession value from the Azul response.
- * @returns Trimmed, non-empty accession IDs.
- */
-function splitAccessionIds(accession: string): string[] {
-  return accession
-    .split(";")
-    .map((id) => id.trim())
-    .filter(Boolean);
+  return buildProjectJsonLd(data, browserURL, OPTIONS);
 }

app/utils/schemaOrg/lungmapProjectDataset.ts

@@ -0,0 +1,27 @@
+import type { ProjectsResponse } from "../../apis/azul/hca-dcp/common/responses";
+import type { ProjectCatalogOptions } from "./projectDataset";
+import { buildProjectJsonLd } from "./projectDataset";
+import type { SchemaDataset } from "./types";
+
+const CATALOG_NAME = "LungMAP Data Explorer";
+
+const OPTIONS: ProjectCatalogOptions = {
+  catalogName: CATALOG_NAME,
+  descriptionFallbackSuffix: `A project in the LungMAP Data Explorer.`,
+};
+
+/**
+ * Builds a Schema.org Dataset JSON-LD object for a LungMAP project. LungMAP
+ * shares the HCA Azul backend, so the response shape matches HCA's
+ * `ProjectsResponse` and the shared `buildProjectJsonLd` core does the
+ * mapping; this wrapper just supplies LungMAP-specific catalog identity.
+ * @param data - LungMAP project detail response from Azul.
+ * @param browserURL - Site base URL used for canonical and catalog URLs.
+ * @returns Schema.org Dataset JSON-LD object, or `undefined` if not buildable.
+ */
+export function buildLungmapProjectJsonLd(
+  data: ProjectsResponse,
+  browserURL: string
+): SchemaDataset | undefined {
+  return buildProjectJsonLd(data, browserURL, OPTIONS);
+}