feat: link species name to ncbi taxon page (#82) (#110)

* feat: link species name to ncbi taxon page (#82)

* fix: uri-encode taxonomy id in ncbi url (#82)

Author: hunterckx

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

@@ -4,6 +4,7 @@ export interface BRCDataCatalogGenome {
   chromosomes: number;
   contigs: number;
   genomeVersionAssemblyId: string;
+  ncbiTaxonomyId: string;
   organism: string;
   species: string;
   strain: string;

app/viewModelBuilders/catalog/brc-analytics-catalog/common/viewModelBuilders.ts

@@ -124,7 +124,15 @@ export const buildGenomeDetails = (
   genome: BRCDataCatalogGenome
 ): ComponentProps<typeof C.KeyValuePairs> => {
   const keyValuePairs = new Map<Key, Value>();
-  keyValuePairs.set("Species", genome.species);
+  keyValuePairs.set(
+    "Species",
+    C.Link({
+      label: genome.species,
+      url: `https://www.ncbi.nlm.nih.gov/datasets/taxonomy/${encodeURIComponent(
+        genome.ncbiTaxonomyId
+      )}/`,
+    })
+  );
   keyValuePairs.set("Strain", genome.strain);
   keyValuePairs.set("Assembly Version ID", genome.genomeVersionAssemblyId);
   keyValuePairs.set("VeUPathDB Project", genome.vEuPathDbProject);
@@ -173,9 +181,12 @@ export const buildOrganismListHero = (): ComponentProps<
  */
 export const buildSpecies = (
   genome: BRCDataCatalogGenome
-): ComponentProps<typeof C.BasicCell> => {
+): ComponentProps<typeof C.Link> => {
   return {
-    value: genome.species,
+    label: genome.species,
+    url: `https://www.ncbi.nlm.nih.gov/datasets/taxonomy/${encodeURIComponent(
+      genome.ncbiTaxonomyId
+    )}/`,
   };
 };
 

files/build-catalog.ts

@@ -23,6 +23,7 @@ async function buildGenomes(): Promise<BRCDataCatalogGenome[]> {
       chromosomes: parseNumber(row.Chromosomes),
       contigs: parseNumber(row.Contigs),
       genomeVersionAssemblyId: row["Genome Version/Assembly ID"],
+      ncbiTaxonomyId: row.taxId,
       organism: row.Organism,
       species: row.Species,
       strain: row.Strain,

files/build-genomes-files.py

@@ -34,6 +34,8 @@ def build_genomes_files():
   if (len(unmatched_assemblies) != 0):
     print(f"Omitted {len(unmatched_assemblies)} assemblies that had no matches: {", ".join(unmatched_assemblies)}")
 
+  result_df["taxId"] = result_df["taxId"].astype(int)
+
   result_df.to_csv(OUTPUT_PATH, index=False, sep="\t")
 
   print(f"Wrote to {OUTPUT_PATH}")