feat: exclude rows without browser urls from catalog input (#17)

galaxyproject · Sep 3, 2024 · d52ab60 · d52ab60
1 parent 06b3181
commit d52ab60
Show file tree

Hide file tree

Showing 5 changed files with 36 additions and 2,776 deletions.
diff --git a/data-catalog/app/apis/catalog/brc-analytics-catalog/common/entities.ts b/data-catalog/app/apis/catalog/brc-analytics-catalog/common/entities.ts
@@ -6,6 +6,6 @@ export interface BRCDataCatalogGenome {
   species: string;
   strain: string;
   supercontigs: number;
-  ucscBrowserUrl: string | null;
+  ucscBrowserUrl: string;
   vEuPathDbProject: string;
 }
diff --git a/data-catalog/files/build-catalog.ts b/data-catalog/files/build-catalog.ts
@@ -27,7 +27,7 @@ async function buildGenomes(): Promise<BRCDataCatalogGenome[]> {
       species: row.Species,
       strain: row.Strain,
       supercontigs: parseNumber(row.Supercontigs),
-      ucscBrowserUrl: parseStringOrNull(row.ucscBrowser),
+      ucscBrowserUrl: row.ucscBrowser,
       vEuPathDbProject: row["VEuPathDB Project"],
     })
   );
@@ -52,10 +52,6 @@ async function saveJson(filePath: string, data: unknown): Promise<void> {
   await fsp.writeFile(filePath, JSON.stringify(data, undefined, 2) + "\n");
 }
 
-function parseStringOrNull(value: string): string | null {
-  return value || null;
-}
-
 function parseNumber(value: string): number {
   value = value.trim();
   const n = Number(value);

diff --git a/data-catalog/files/build-genomes-files.py b/data-catalog/files/build-genomes-files.py
@@ -16,7 +16,7 @@ def build_genomes_files():
   gen_bank_merge_df = genomes_source_df.merge(assemblies_df, how="left", left_on="Genome Version/Assembly ID", right_on="genBank")
   ref_seq_merge_df = genomes_source_df.merge(assemblies_df, how="left", left_on="Genome Version/Assembly ID", right_on="refSeq")
 
-  result_df = gen_bank_merge_df.combine_first(ref_seq_merge_df)
+  result_df = gen_bank_merge_df.combine_first(ref_seq_merge_df).dropna(subset=["ucscBrowser"])
 
   result_df.to_csv(OUTPUT_PATH, index=False, sep="\t")