feat: add organism list (#177)

galaxyproject · Nov 19, 2024 · 37a5cef · 37a5cef
1 parent 24969d7
commit 37a5cef
Show file tree

Hide file tree

Showing 9 changed files with 316 additions and 16 deletions.
diff --git a/app/apis/catalog/brc-analytics-catalog/common/entities.ts b/app/apis/catalog/brc-analytics-catalog/common/entities.ts
@@ -28,6 +28,13 @@ export interface BRCDataCatalogGenome {
   ucscBrowserUrl: string | null;
 }
 
+export interface BRCDataCatalogOrganism {
+  assemblyCount: number;
+  ncbiTaxonomyId: string;
+  tags: string[];
+  taxon: string;
+}
+
 export interface EntitiesResponse<R> {
   hits: R[];
   pagination: EntitiesResponsePagination;

diff --git a/app/apis/catalog/brc-analytics-catalog/common/utils.ts b/app/apis/catalog/brc-analytics-catalog/common/utils.ts
@@ -1,4 +1,4 @@
-import { BRCDataCatalogGenome } from "./entities";
+import { BRCDataCatalogGenome, BRCDataCatalogOrganism } from "./entities";
 
 export function getGenomeId(genome: BRCDataCatalogGenome): string {
   return sanitizeEntityId(genome.accession);
@@ -9,6 +9,10 @@ export function getGenomeTitle(genome?: BRCDataCatalogGenome): string {
   return `${genome.taxon}`;
 }
 
+export function getOrganismId(organism: BRCDataCatalogOrganism): string {
+  return sanitizeEntityId(organism.ncbiTaxonomyId);
+}
+
 export function sanitizeEntityId(entityId?: string): string {
   if (!entityId) return "";
   return entityId.replace(/\./g, "_");

diff --git a/app/viewModelBuilders/catalog/brc-analytics-catalog/common/viewModelBuilders.ts b/app/viewModelBuilders/catalog/brc-analytics-catalog/common/viewModelBuilders.ts
@@ -6,7 +6,10 @@ import {
 import { ViewContext } from "@databiosphere/findable-ui/lib/config/entities";
 import { ComponentProps } from "react";
 import { ROUTES } from "../../../../../routes/constants";
-import { BRCDataCatalogGenome } from "../../../../apis/catalog/brc-analytics-catalog/common/entities";
+import {
+  BRCDataCatalogGenome,
+  BRCDataCatalogOrganism,
+} from "../../../../apis/catalog/brc-analytics-catalog/common/entities";
 import * as C from "../../../../components";
 import { GENOME_BROWSER, NCBI_DATASETS_URL } from "./constants";
 
@@ -71,6 +74,19 @@ export const buildAnnotationStatus = (
   };
 };
 
+/**
+ * Build props for the assemblies cell.
+ * @param organism - Genome entity.
+ * @returns Props to be used for the cell.
+ */
+export const buildAssemblyCount = (
+  organism: BRCDataCatalogOrganism
+): ComponentProps<typeof C.BasicCell> => {
+  return {
+    value: organism.assemblyCount,
+  };
+};
+
 /**
  * Build props for the chromosomes cell.
  * @param genome - Genome entity.
@@ -155,7 +171,7 @@ export const buildLevel = (
  * @returns Props to be used for the cell.
  */
 export const buildTaxon = (
-  genome: BRCDataCatalogGenome
+  genome: BRCDataCatalogOrganism | BRCDataCatalogGenome
 ): ComponentProps<typeof C.BasicCell> => {
   return {
     value: genome.taxon,
@@ -207,7 +223,7 @@ export const buildScaffoldN50 = (
  * @returns Props to be used for the cell.
  */
 export const buildTags = (
-  genome: BRCDataCatalogGenome
+  genome: BRCDataCatalogOrganism | BRCDataCatalogGenome
 ): ComponentProps<typeof C.NTagCell> => {
   return {
     label: "Tags",
@@ -221,7 +237,7 @@ export const buildTags = (
  * @returns Props to be used for the cell.
  */
 export const buildTaxonomyId = (
-  genome: BRCDataCatalogGenome
+  genome: BRCDataCatalogOrganism | BRCDataCatalogGenome
 ): ComponentProps<typeof C.BasicCell> => {
   return {
     value: genome.ncbiTaxonomyId,

diff --git a/files/build-catalog.ts b/files/build-catalog.ts
@@ -1,6 +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 {
+  BRCDataCatalogGenome,
+  BRCDataCatalogOrganism,
+} from "../app/apis/catalog/brc-analytics-catalog/common/entities";
 import { SourceGenome, SourceOrganism } from "./entities";
 
 const SOURCE_PATH_ORGANISMS = "files/source/organisms-from-ncbi.tsv";
@@ -9,24 +12,45 @@ const SOURCE_PATH_GENOMES = "files/source/genomes-from-ncbi.tsv";
 buildCatalog();
 
 async function buildCatalog(): Promise<void> {
-  const genomes = await buildGenomes();
+  const organisms = await buildOrganisms();
+
+  const organismsByTaxon = new Map(
+    organisms.map((organism) => [organism.taxon, organism])
+  );
+
+  const genomes = await buildGenomes(organismsByTaxon);
+
+  console.log("Organisms:", genomes.length);
+  await saveJson("files/out/organisms.json", organisms);
 
   console.log("Genomes:", genomes.length);
   await saveJson("files/out/genomes.json", genomes);
 
   console.log("Done");
 }
 
-async function buildGenomes(): Promise<BRCDataCatalogGenome[]> {
-  const sourceOrganismRows = await readValuesFile<SourceOrganism>(
+async function buildOrganisms(): Promise<BRCDataCatalogOrganism[]> {
+  const sourceRows = await readValuesFile<SourceOrganism>(
     SOURCE_PATH_ORGANISMS
   );
-  const sourceOrganismsByTaxon = new Map(
-    sourceOrganismRows.map((row) => [row.taxon, row])
+  const mappedRows = sourceRows.map((row): BRCDataCatalogOrganism => {
+    return {
+      assemblyCount: parseNumber(row.assemblyCount),
+      ncbiTaxonomyId: row.taxonomyId,
+      tags: row.CustomTags ? [row.CustomTags] : [],
+      taxon: row.taxon,
+    };
+  });
+  return mappedRows.sort((a, b) =>
+    a.ncbiTaxonomyId.localeCompare(b.ncbiTaxonomyId)
   );
+}
+
+async function buildGenomes(
+  organismsByTaxon: Map<string, BRCDataCatalogOrganism>
+): Promise<BRCDataCatalogGenome[]> {
   const sourceRows = await readValuesFile<SourceGenome>(SOURCE_PATH_GENOMES);
   const mappedRows = sourceRows.map((row): BRCDataCatalogGenome => {
-    const tagsString = sourceOrganismsByTaxon.get(row.taxon)?.CustomTags;
     return {
       accession: row.accession,
       annotationStatus: parseStringOrNull(row.annotationStatus),
@@ -41,7 +65,7 @@ async function buildGenomes(): Promise<BRCDataCatalogGenome[]> {
       scaffoldCount: parseNumber(row.scaffoldCount),
       scaffoldL50: parseNumber(row.scaffoldL50),
       scaffoldN50: parseNumber(row.scaffoldN50),
-      tags: tagsString ? [tagsString] : [],
+      tags: organismsByTaxon.get(row.taxon)?.tags ?? [],
       taxon: row.taxon,
       ucscBrowserUrl: parseStringOrNull(row.ucscBrowser),
     };

diff --git a/files/out/organisms.json b/files/out/organisms.json
@@ -0,0 +1,136 @@
+[
+  {
+    "assemblyCount": 6911,
+    "ncbiTaxonomyId": "10244",
+    "tags": [
+      "Virus"
+    ],
+    "taxon": "Monkeypox virus"
+  },
+  {
+    "assemblyCount": 7823,
+    "ncbiTaxonomyId": "1773",
+    "tags": [
+      "Bact"
+    ],
+    "taxon": "Mycobacterium tuberculosis"
+  },
+  {
+    "assemblyCount": 13,
+    "ncbiTaxonomyId": "199306",
+    "tags": [
+      "VEuPathDb"
+    ],
+    "taxon": "Coccidioides posadasii"
+  },
+  {
+    "assemblyCount": 92,
+    "ncbiTaxonomyId": "2697049",
+    "tags": [
+      "Virus"
+    ],
+    "taxon": "Severe acute respiratory syndrome coronavirus 2"
+  },
+  {
+    "assemblyCount": 5,
+    "ncbiTaxonomyId": "5501",
+    "tags": [],
+    "taxon": "Coccidioides immitis"
+  },
+  {
+    "assemblyCount": 11,
+    "ncbiTaxonomyId": "5660",
+    "tags": [
+      "VEuPathDb"
+    ],
+    "taxon": "Leishmania braziliensis"
+  },
+  {
+    "assemblyCount": 12,
+    "ncbiTaxonomyId": "5661",
+    "tags": [
+      "VEuPathDb"
+    ],
+    "taxon": "Leishmania donovani"
+  },
+  {
+    "assemblyCount": 7,
+    "ncbiTaxonomyId": "5664",
+    "tags": [
+      "VEuPathDb"
+    ],
+    "taxon": "Leishmania major"
+  },
+  {
+    "assemblyCount": 5,
+    "ncbiTaxonomyId": "5691",
+    "tags": [
+      "VEuPathDb"
+    ],
+    "taxon": "Trypanosoma brucei"
+  },
+  {
+    "assemblyCount": 44,
+    "ncbiTaxonomyId": "5693",
+    "tags": [
+      "VEuPathDb"
+    ],
+    "taxon": "Trypanosoma cruzi"
+  },
+  {
+    "assemblyCount": 29,
+    "ncbiTaxonomyId": "5811",
+    "tags": [
+      "VEuPathDb"
+    ],
+    "taxon": "Toxoplasma gondii"
+  },
+  {
+    "assemblyCount": 67,
+    "ncbiTaxonomyId": "5833",
+    "tags": [
+      "VEuPathDb"
+    ],
+    "taxon": "Plasmodium falciparum"
+  },
+  {
+    "assemblyCount": 19,
+    "ncbiTaxonomyId": "5855",
+    "tags": [
+      "VEuPathDb"
+    ],
+    "taxon": "Plasmodium vivax"
+  },
+  {
+    "assemblyCount": 10,
+    "ncbiTaxonomyId": "5860",
+    "tags": [
+      "VEuPathDb"
+    ],
+    "taxon": "Plasmodium vinckei"
+  },
+  {
+    "assemblyCount": 15,
+    "ncbiTaxonomyId": "5861",
+    "tags": [
+      "VEuPathDb"
+    ],
+    "taxon": "Plasmodium yoelii"
+  },
+  {
+    "assemblyCount": 7,
+    "ncbiTaxonomyId": "7165",
+    "tags": [
+      "VEuPathDb"
+    ],
+    "taxon": "Anopheles gambiae"
+  },
+  {
+    "assemblyCount": 5,
+    "ncbiTaxonomyId": "7175",
+    "tags": [
+      "VEuPathDb"
+    ],
+    "taxon": "Culex pipiens"
+  }
+]
diff --git a/routes/constants.ts b/routes/constants.ts
@@ -1,5 +1,6 @@
 export const ROUTES = {
   ABOUT: "/about",
   GENOMES: "/data/genomes",
+  ORGANISMS: "/data/organisms",
   ROADMAP: "/roadmap",
 };
diff --git a/site-config/brc-analytics/category.ts b/site-config/brc-analytics/category.ts
@@ -2,6 +2,7 @@ export const BRC_DATA_CATALOG_CATEGORY_KEY = {
   ACCESSION: "accession",
   ANALYZE_GENOME: "analyzeGenome",
   ANNOTATION_STATUS: "annotationStatus",
+  ASSEMBLY_COUNT: "assemblyCount",
   CHROMOSOMES: "chromosomes",
   COVERAGE: "coverage",
   GC_PERCENT: "gcPercent",
@@ -21,6 +22,7 @@ export const BRC_DATA_CATALOG_CATEGORY_LABEL = {
   ACCESSION: "Accession",
   ANALYZE_GENOME: "Action",
   ANNOTATION_STATUS: "Annotation Status",
+  ASSEMBLY_COUNT: "Assemblies",
   CHROMOSOMES: "Chromosomes",
   COVERAGE: "Coverage",
   GC_PERCENT: "GC%",

diff --git a/site-config/brc-analytics/local/config.ts b/site-config/brc-analytics/local/config.ts
@@ -1,10 +1,14 @@
 import { SiteConfig } from "@databiosphere/findable-ui/lib/config/entities";
 import { EntityConfig } from "@databiosphere/findable-ui/src/config/entities";
-import { BRCDataCatalogGenome } from "../../../app/apis/catalog/brc-analytics-catalog/common/entities";
+import {
+  BRCDataCatalogGenome,
+  BRCDataCatalogOrganism,
+} from "../../../app/apis/catalog/brc-analytics-catalog/common/entities";
 import * as C from "../../../app/components";
 import { ROUTES } from "../../../routes/constants";
 import { floating } from "./floating/floating";
 import { genomeEntityConfig } from "./index/genomeEntityConfig";
+import { organismEntityConfig } from "./index/organismEntityConfig";
 
 const LOCALHOST = "http://localhost:3000";
 const APP_TITLE = "BRC Analytics";
@@ -32,7 +36,10 @@ export function makeConfig(browserUrl: string): SiteConfig {
     dataSource: {
       url: "",
     },
-    entities: [genomeEntityConfig as EntityConfig<BRCDataCatalogGenome>],
+    entities: [
+      organismEntityConfig as EntityConfig<BRCDataCatalogOrganism>,
+      genomeEntityConfig as EntityConfig<BRCDataCatalogGenome>,
+    ],
     explorerTitle: APP_TITLE,
     layout: {
       floating,
@@ -50,7 +57,8 @@ export function makeConfig(browserUrl: string): SiteConfig {
           undefined,
           [
             { label: "About", url: ROUTES.ABOUT },
-            { label: "Datasets", url: ROUTES.GENOMES },
+            { label: "Organisms", url: ROUTES.ORGANISMS },
+            { label: "Genomes", url: ROUTES.GENOMES },
             { label: "Roadmap", url: ROUTES.ROADMAP },
           ],
           undefined,