feat: add tags to genome list (#177)

galaxyproject · Nov 18, 2024 · 24969d7 · 24969d7
1 parent 910ffd2
commit 24969d7
Show file tree

Hide file tree

Showing 8 changed files with 90 additions and 5 deletions.
diff --git a/app/apis/catalog/brc-analytics-catalog/common/entities.ts b/app/apis/catalog/brc-analytics-catalog/common/entities.ts
@@ -23,6 +23,7 @@ export interface BRCDataCatalogGenome {
   scaffoldCount: number;
   scaffoldL50: number;
   scaffoldN50: number;
+  tags: string[];
   taxon: string;
   ucscBrowserUrl: string | null;
 }

diff --git a/app/components/index.ts b/app/components/index.ts
@@ -21,6 +21,7 @@ export {
 export { Logo } from "@databiosphere/findable-ui/lib/components/Layout/components/Header/components/Content/components/Logo/logo";
 export { Link } from "@databiosphere/findable-ui/lib/components/Links/components/Link/link";
 export { BasicCell } from "@databiosphere/findable-ui/lib/components/Table/components/TableCell/components/BasicCell/basicCell";
+export { NTagCell } from "@databiosphere/findable-ui/lib/components/Table/components/TableCell/components/NTagCell/nTagCell";
 export { CopyText } from "./common/CopyText/copyText";
 export { AnalysisMethod } from "./Entity/components/AnalysisMethod/analysisMethod";
 export { AnalysisMethods } from "./Entity/components/AnalysisMethods/analysisMethods";

diff --git a/app/viewModelBuilders/catalog/brc-analytics-catalog/common/viewModelBuilders.ts b/app/viewModelBuilders/catalog/brc-analytics-catalog/common/viewModelBuilders.ts
@@ -201,6 +201,20 @@ export const buildScaffoldN50 = (
   };
 };
 
+/**
+ * Build props for the tags cell.
+ * @param genome - Genome entity.
+ * @returns Props to be used for the cell.
+ */
+export const buildTags = (
+  genome: BRCDataCatalogGenome
+): ComponentProps<typeof C.NTagCell> => {
+  return {
+    label: "Tags",
+    values: genome.tags,
+  };
+};
+
 /**
  * Build props for the taxonomy ID cell.
  * @param genome - Genome entity.

diff --git a/files/build-catalog.ts b/files/build-catalog.ts
@@ -1,8 +1,9 @@
 import { parse as parseCsv } from "csv-parse/sync";
 import fsp from "fs/promises";
 import { BRCDataCatalogGenome } from "../app/apis/catalog/brc-analytics-catalog/common/entities";
-import { SourceGenome } from "./entities";
+import { SourceGenome, SourceOrganism } from "./entities";
 
+const SOURCE_PATH_ORGANISMS = "files/source/organisms-from-ncbi.tsv";
 const SOURCE_PATH_GENOMES = "files/source/genomes-from-ncbi.tsv";
 
 buildCatalog();
@@ -17,9 +18,16 @@ async function buildCatalog(): Promise<void> {
 }
 
 async function buildGenomes(): Promise<BRCDataCatalogGenome[]> {
+  const sourceOrganismRows = await readValuesFile<SourceOrganism>(
+    SOURCE_PATH_ORGANISMS
+  );
+  const sourceOrganismsByTaxon = new Map(
+    sourceOrganismRows.map((row) => [row.taxon, row])
+  );
   const sourceRows = await readValuesFile<SourceGenome>(SOURCE_PATH_GENOMES);
-  const mappedRows = sourceRows.map(
-    (row): BRCDataCatalogGenome => ({
+  const mappedRows = sourceRows.map((row): BRCDataCatalogGenome => {
+    const tagsString = sourceOrganismsByTaxon.get(row.taxon)?.CustomTags;
+    return {
       accession: row.accession,
       annotationStatus: parseStringOrNull(row.annotationStatus),
       chromosomes: parseNumberOrNull(row.chromosomeCount),
@@ -33,10 +41,11 @@ async function buildGenomes(): Promise<BRCDataCatalogGenome[]> {
       scaffoldCount: parseNumber(row.scaffoldCount),
       scaffoldL50: parseNumber(row.scaffoldL50),
       scaffoldN50: parseNumber(row.scaffoldN50),
+      tags: tagsString ? [tagsString] : [],
       taxon: row.taxon,
       ucscBrowserUrl: parseStringOrNull(row.ucscBrowser),
-    })
-  );
+    };
+  });
   return mappedRows.sort((a, b) => a.accession.localeCompare(b.accession));
 }
 

diff --git a/files/entities.ts b/files/entities.ts
@@ -15,3 +15,10 @@ export interface SourceGenome {
   taxonomyId: string;
   ucscBrowser: string;
 }
+
+export interface SourceOrganism {
+  assemblyCount: string;
+  CustomTags: string;
+  taxon: string;
+  taxonomyId: string;
+}
diff --git a/files/out/genomes.json b/files/out/genomes.json
@@ -13,6 +13,9 @@
     "scaffoldCount": 2747,
     "scaffoldL50": 6,
     "scaffoldN50": 1678596,
+    "tags": [
+      "VEuPathDb"
+    ],
     "taxon": "Plasmodium vivax",
     "ucscBrowserUrl": "https://genome.ucsc.edu/h/GCF_000002415.2"
   },
@@ -30,6 +33,7 @@
     "scaffoldCount": 12,
     "scaffoldL50": 4,
     "scaffoldN50": 2481190,
+    "tags": [],
     "taxon": "Trypanosoma brucei brucei TREU927",
     "ucscBrowserUrl": "https://genome.ucsc.edu/h/GCF_000002445.2"
   },
@@ -47,6 +51,7 @@
     "scaffoldCount": 36,
     "scaffoldL50": 11,
     "scaffoldN50": 1091540,
+    "tags": [],
     "taxon": "Leishmania major strain Friedlin",
     "ucscBrowserUrl": "https://genome.ucsc.edu/h/GCF_000002725.2"
   },
@@ -64,6 +69,7 @@
     "scaffoldCount": 14,
     "scaffoldL50": 5,
     "scaffoldN50": 1687656,
+    "tags": [],
     "taxon": "Plasmodium falciparum 3D7",
     "ucscBrowserUrl": "https://genome.ucsc.edu/h/GCF_000002765.5"
   },
@@ -81,6 +87,7 @@
     "scaffoldCount": 138,
     "scaffoldL50": 11,
     "scaffoldN50": 992961,
+    "tags": [],
     "taxon": "Leishmania braziliensis MHOM/BR/75/M2904",
     "ucscBrowserUrl": "https://genome.ucsc.edu/h/GCF_000002845.2"
   },
@@ -98,6 +105,7 @@
     "scaffoldCount": 2276,
     "scaffoldL50": 6,
     "scaffoldN50": 4973582,
+    "tags": [],
     "taxon": "Toxoplasma gondii ME49",
     "ucscBrowserUrl": "https://genome.ucsc.edu/h/GCF_000006565.2"
   },
@@ -115,6 +123,7 @@
     "scaffoldCount": 6,
     "scaffoldL50": 3,
     "scaffoldN50": 4323945,
+    "tags": [],
     "taxon": "Coccidioides immitis RS",
     "ucscBrowserUrl": "https://genome.ucsc.edu/h/GCF_000149335.2"
   },
@@ -132,6 +141,7 @@
     "scaffoldCount": 1,
     "scaffoldL50": 1,
     "scaffoldN50": 4411532,
+    "tags": [],
     "taxon": "Mycobacterium tuberculosis H37Rv",
     "ucscBrowserUrl": null
   },
@@ -149,6 +159,9 @@
     "scaffoldCount": 29495,
     "scaffoldL50": 212,
     "scaffoldN50": 88624,
+    "tags": [
+      "VEuPathDb"
+    ],
     "taxon": "Trypanosoma cruzi",
     "ucscBrowserUrl": "https://genome.ucsc.edu/h/GCF_000209065.1"
   },
@@ -166,6 +179,9 @@
     "scaffoldCount": 36,
     "scaffoldL50": 11,
     "scaffoldN50": 1024085,
+    "tags": [
+      "VEuPathDb"
+    ],
     "taxon": "Leishmania donovani",
     "ucscBrowserUrl": "https://genome.ucsc.edu/h/GCF_000227135.1"
   },
@@ -183,6 +199,7 @@
     "scaffoldCount": 1,
     "scaffoldL50": 1,
     "scaffoldN50": 4411709,
+    "tags": [],
     "taxon": "Mycobacterium tuberculosis H37Rv",
     "ucscBrowserUrl": null
   },
@@ -200,6 +217,9 @@
     "scaffoldCount": 1,
     "scaffoldL50": 1,
     "scaffoldN50": 196858,
+    "tags": [
+      "Virus"
+    ],
     "taxon": "Monkeypox virus",
     "ucscBrowserUrl": null
   },
@@ -217,6 +237,9 @@
     "scaffoldCount": 1,
     "scaffoldL50": 1,
     "scaffoldN50": 29903,
+    "tags": [
+      "Virus"
+    ],
     "taxon": "Severe acute respiratory syndrome coronavirus 2",
     "ucscBrowserUrl": null
   },
@@ -234,6 +257,7 @@
     "scaffoldCount": 289,
     "scaffoldL50": 2,
     "scaffoldN50": 186194774,
+    "tags": [],
     "taxon": "Culex pipiens pallens",
     "ucscBrowserUrl": null
   },
@@ -251,6 +275,7 @@
     "scaffoldCount": 9,
     "scaffoldL50": 2,
     "scaffoldN50": 8079863,
+    "tags": [],
     "taxon": "Coccidioides posadasii str. Silveira",
     "ucscBrowserUrl": "https://genome.ucsc.edu/h/GCA_018416015.2"
   },
@@ -268,6 +293,9 @@
     "scaffoldCount": 1,
     "scaffoldL50": 1,
     "scaffoldN50": 4516435,
+    "tags": [
+      "Bact"
+    ],
     "taxon": "Mycobacterium tuberculosis",
     "ucscBrowserUrl": null
   },
@@ -285,6 +313,9 @@
     "scaffoldCount": 14,
     "scaffoldL50": 5,
     "scaffoldN50": 2046250,
+    "tags": [
+      "VEuPathDb"
+    ],
     "taxon": "Plasmodium yoelii",
     "ucscBrowserUrl": "https://genome.ucsc.edu/h/GCF_900002385.2"
   },
@@ -302,6 +333,7 @@
     "scaffoldCount": 14,
     "scaffoldL50": 5,
     "scaffoldN50": 1692345,
+    "tags": [],
     "taxon": "Plasmodium vinckei vinckei",
     "ucscBrowserUrl": "https://genome.ucsc.edu/h/GCF_900681995.1"
   },
@@ -319,6 +351,9 @@
     "scaffoldCount": 190,
     "scaffoldL50": 2,
     "scaffoldN50": 99149756,
+    "tags": [
+      "VEuPathDb"
+    ],
     "taxon": "Anopheles gambiae",
     "ucscBrowserUrl": "https://genome.ucsc.edu/h/GCF_943734735.2"
   },
@@ -336,6 +371,9 @@
     "scaffoldCount": 1,
     "scaffoldL50": 1,
     "scaffoldN50": 4469156,
+    "tags": [
+      "Bact"
+    ],
     "taxon": "Mycobacterium tuberculosis",
     "ucscBrowserUrl": null
   }

diff --git a/site-config/brc-analytics/category.ts b/site-config/brc-analytics/category.ts
@@ -11,6 +11,7 @@ export const BRC_DATA_CATALOG_CATEGORY_KEY = {
   SCAFFOLD_COUNT: "scaffoldCount",
   SCAFFOLD_L50: "scaffoldL50",
   SCAFFOLD_N50: "scaffoldN50",
+  TAGS: "tags",
   TAXON: "taxon",
   TAXONOMY_ID: "ncbiTaxonomyId",
   UCSC_BROWSER_URL: "ucscBrowserUrl",
@@ -29,6 +30,7 @@ export const BRC_DATA_CATALOG_CATEGORY_LABEL = {
   SCAFFOLD_COUNT: "Scaffolds",
   SCAFFOLD_L50: "Scaffold L50",
   SCAFFOLD_N50: "Scaffold N50",
+  TAGS: "Tags",
   TAXON: "Taxon",
   TAXONOMY_ID: "Taxonomy ID",
   UCSC_BROWSER_URL: "UCSC Browser",

diff --git a/site-config/brc-analytics/local/index/genomeEntityConfig.ts b/site-config/brc-analytics/local/index/genomeEntityConfig.ts
@@ -61,6 +61,10 @@ export const genomeEntityConfig: BRCEntityConfig<BRCDataCatalogGenome> = {
             key: BRC_DATA_CATALOG_CATEGORY_KEY.ANNOTATION_STATUS,
             label: BRC_DATA_CATALOG_CATEGORY_LABEL.ANNOTATION_STATUS,
           },
+          {
+            key: BRC_DATA_CATALOG_CATEGORY_KEY.TAGS,
+            label: BRC_DATA_CATALOG_CATEGORY_LABEL.TAGS,
+          },
         ],
       },
     ],
@@ -214,6 +218,15 @@ export const genomeEntityConfig: BRCEntityConfig<BRCDataCatalogGenome> = {
         id: BRC_DATA_CATALOG_CATEGORY_KEY.ANNOTATION_STATUS,
         width: { max: "0.5fr", min: "142px" },
       },
+      {
+        componentConfig: {
+          component: C.NTagCell,
+          viewBuilder: V.buildTags,
+        } as ComponentConfig<typeof C.NTagCell, BRCDataCatalogGenome>,
+        header: BRC_DATA_CATALOG_CATEGORY_LABEL.TAGS,
+        id: BRC_DATA_CATALOG_CATEGORY_KEY.TAGS,
+        width: { max: "0.5fr", min: "142px" },
+      },
     ],
     defaultSort: {
       desc: SORT_DIRECTION.ASCENDING,