feat: use assembly taxonomy ids for organism taxonomy id column (#210)

galaxyproject · Jan 16, 2025 · 3ca8188 · 3ca8188
1 parent 6990ba7
commit 3ca8188
Show file tree

Hide file tree

Showing 6 changed files with 92 additions and 7 deletions.
diff --git a/app/apis/catalog/brc-analytics-catalog/common/entities.ts b/app/apis/catalog/brc-analytics-catalog/common/entities.ts
@@ -32,6 +32,7 @@ export interface BRCDataCatalogGenome {
 
 export interface BRCDataCatalogOrganism {
   assemblyCount: number;
+  assemblyTaxonomyIds: string[];
   genomes: BRCDataCatalogGenome[];
   ncbiTaxonomyId: string;
   species: string;

diff --git a/app/viewModelBuilders/catalog/brc-analytics-catalog/common/viewModelBuilders.ts b/app/viewModelBuilders/catalog/brc-analytics-catalog/common/viewModelBuilders.ts
@@ -184,6 +184,20 @@ export const buildLevel = (
   };
 };
 
+/**
+ * Build props for the assembly taxonomy IDs cell.
+ * @param organism - Organism entity.
+ * @returns Props to be used for the cell.
+ */
+export const buildOrganismAssemblyTaxonomyIds = (
+  organism: BRCDataCatalogOrganism
+): ComponentProps<typeof C.NTagCell> => {
+  return {
+    label: "taxonomy IDs",
+    values: organism.assemblyTaxonomyIds,
+  };
+};
+
 /**
  * Build props for the taxon cell.
  * @param organism - Organism entity.

diff --git a/files/build-catalog.ts b/files/build-catalog.ts
@@ -71,6 +71,9 @@ function buildOrganism(
 ): BRCDataCatalogOrganism {
   return {
     assemblyCount: (organism?.assemblyCount ?? 0) + 1,
+    assemblyTaxonomyIds: Array.from(
+      new Set([...(organism?.assemblyTaxonomyIds ?? []), genome.ncbiTaxonomyId])
+    ),
     genomes: [...(organism?.genomes ?? []), genome],
     ncbiTaxonomyId: genome.speciesTaxonomyId,
     species: genome.species,

diff --git a/files/out/organisms.json b/files/out/organisms.json
@@ -1,6 +1,9 @@
 [
   {
     "assemblyCount": 1,
+    "assemblyTaxonomyIds": [
+      "10244"
+    ],
     "genomes": [
       {
         "accession": "GCF_000857045.1",
@@ -35,6 +38,9 @@
   },
   {
     "assemblyCount": 1,
+    "assemblyTaxonomyIds": [
+      "83332"
+    ],
     "genomes": [
       {
         "accession": "GCF_000195955.2",
@@ -69,6 +75,9 @@
   },
   {
     "assemblyCount": 2,
+    "assemblyTaxonomyIds": [
+      "443226"
+    ],
     "genomes": [
       {
         "accession": "GCA_000170175.2",
@@ -124,6 +133,10 @@
   },
   {
     "assemblyCount": 2,
+    "assemblyTaxonomyIds": [
+      "235443",
+      "214684"
+    ],
     "genomes": [
       {
         "accession": "GCA_000149245.3",
@@ -179,6 +192,9 @@
   },
   {
     "assemblyCount": 1,
+    "assemblyTaxonomyIds": [
+      "237561"
+    ],
     "genomes": [
       {
         "accession": "GCF_000182965.3",
@@ -211,6 +227,9 @@
   },
   {
     "assemblyCount": 1,
+    "assemblyTaxonomyIds": [
+      "246410"
+    ],
     "genomes": [
       {
         "accession": "GCF_000149335.2",
@@ -243,6 +262,9 @@
   },
   {
     "assemblyCount": 1,
+    "assemblyTaxonomyIds": [
+      "420245"
+    ],
     "genomes": [
       {
         "accession": "GCF_000002845.2",
@@ -275,6 +297,9 @@
   },
   {
     "assemblyCount": 1,
+    "assemblyTaxonomyIds": [
+      "5661"
+    ],
     "genomes": [
       {
         "accession": "GCF_000227135.1",
@@ -307,6 +332,9 @@
   },
   {
     "assemblyCount": 1,
+    "assemblyTaxonomyIds": [
+      "347515"
+    ],
     "genomes": [
       {
         "accession": "GCF_000002725.2",
@@ -339,6 +367,9 @@
   },
   {
     "assemblyCount": 1,
+    "assemblyTaxonomyIds": [
+      "185431"
+    ],
     "genomes": [
       {
         "accession": "GCF_000002445.2",
@@ -371,6 +402,9 @@
   },
   {
     "assemblyCount": 1,
+    "assemblyTaxonomyIds": [
+      "5693"
+    ],
     "genomes": [
       {
         "accession": "GCF_000209065.1",
@@ -403,6 +437,9 @@
   },
   {
     "assemblyCount": 1,
+    "assemblyTaxonomyIds": [
+      "508771"
+    ],
     "genomes": [
       {
         "accession": "GCF_000006565.2",
@@ -435,6 +472,9 @@
   },
   {
     "assemblyCount": 1,
+    "assemblyTaxonomyIds": [
+      "36329"
+    ],
     "genomes": [
       {
         "accession": "GCF_000002765.6",
@@ -467,6 +507,9 @@
   },
   {
     "assemblyCount": 2,
+    "assemblyTaxonomyIds": [
+      "5855"
+    ],
     "genomes": [
       {
         "accession": "GCA_900093555.2",
@@ -522,6 +565,9 @@
   },
   {
     "assemblyCount": 1,
+    "assemblyTaxonomyIds": [
+      "54757"
+    ],
     "genomes": [
       {
         "accession": "GCF_900681995.1",
@@ -554,6 +600,9 @@
   },
   {
     "assemblyCount": 1,
+    "assemblyTaxonomyIds": [
+      "5861"
+    ],
     "genomes": [
       {
         "accession": "GCF_900002385.2",
@@ -586,6 +635,9 @@
   },
   {
     "assemblyCount": 1,
+    "assemblyTaxonomyIds": [
+      "2697049"
+    ],
     "genomes": [
       {
         "accession": "GCF_009858895.2",
@@ -620,6 +672,10 @@
   },
   {
     "assemblyCount": 2,
+    "assemblyTaxonomyIds": [
+      "180454",
+      "7165"
+    ],
     "genomes": [
       {
         "accession": "GCF_000005575.2",
@@ -675,6 +731,9 @@
   },
   {
     "assemblyCount": 1,
+    "assemblyTaxonomyIds": [
+      "42434"
+    ],
     "genomes": [
       {
         "accession": "GCF_016801865.2",
@@ -707,6 +766,9 @@
   },
   {
     "assemblyCount": 1,
+    "assemblyTaxonomyIds": [
+      "7176"
+    ],
     "genomes": [
       {
         "accession": "GCF_015732765.1",
@@ -739,6 +801,9 @@
   },
   {
     "assemblyCount": 1,
+    "assemblyTaxonomyIds": [
+      "330879"
+    ],
     "genomes": [
       {
         "accession": "GCF_000002655.1",

diff --git a/site-config/brc-analytics/category.ts b/site-config/brc-analytics/category.ts
@@ -3,6 +3,7 @@ export const BRC_DATA_CATALOG_CATEGORY_KEY = {
   ANALYZE_GENOME: "analyzeGenome",
   ANNOTATION_STATUS: "annotationStatus",
   ASSEMBLY_COUNT: "assemblyCount",
+  ASSEMBLY_TAXONOMY_IDS: "assemblyTaxonomyIds",
   CHROMOSOMES: "chromosomes",
   COVERAGE: "coverage",
   GC_PERCENT: "gcPercent",
@@ -23,6 +24,7 @@ export const BRC_DATA_CATALOG_CATEGORY_LABEL = {
   ANALYZE_GENOME: "Action",
   ANNOTATION_STATUS: "Annotation Status",
   ASSEMBLY_COUNT: "Assemblies",
+  ASSEMBLY_TAXONOMY_IDS: "Taxonomy ID",
   CHROMOSOMES: "Chromosomes",
   COVERAGE: "Coverage",
   GC_PERCENT: "GC%",

diff --git a/site-config/brc-analytics/local/index/organismEntityConfig.ts b/site-config/brc-analytics/local/index/organismEntityConfig.ts
@@ -29,8 +29,8 @@ export const organismEntityConfig: BRCEntityConfig<BRCDataCatalogOrganism> = {
             label: BRC_DATA_CATALOG_CATEGORY_LABEL.SPECIES,
           },
           {
-            key: BRC_DATA_CATALOG_CATEGORY_KEY.TAXONOMY_ID,
-            label: BRC_DATA_CATALOG_CATEGORY_LABEL.TAXONOMY_ID,
+            key: BRC_DATA_CATALOG_CATEGORY_KEY.ASSEMBLY_TAXONOMY_IDS,
+            label: BRC_DATA_CATALOG_CATEGORY_LABEL.ASSEMBLY_TAXONOMY_IDS,
           },
         ],
       },
@@ -66,11 +66,11 @@ export const organismEntityConfig: BRCEntityConfig<BRCDataCatalogOrganism> = {
       },
       {
         componentConfig: {
-          component: C.BasicCell,
-          viewBuilder: V.buildTaxonomyId,
-        } as ComponentConfig<typeof C.BasicCell, BRCDataCatalogOrganism>,
-        header: BRC_DATA_CATALOG_CATEGORY_LABEL.TAXONOMY_ID,
-        id: BRC_DATA_CATALOG_CATEGORY_KEY.TAXONOMY_ID,
+          component: C.NTagCell,
+          viewBuilder: V.buildOrganismAssemblyTaxonomyIds,
+        } as ComponentConfig<typeof C.NTagCell, BRCDataCatalogOrganism>,
+        header: BRC_DATA_CATALOG_CATEGORY_LABEL.ASSEMBLY_TAXONOMY_IDS,
+        id: BRC_DATA_CATALOG_CATEGORY_KEY.ASSEMBLY_TAXONOMY_IDS,
         width: { max: "0.5fr", min: "164px" },
       },
       {