Merge pull request #20 from biosustain/export_apis

Add apis to fetch a list of genes/strains

Author: SHINTEKI

django_project/urls.py

@@ -218,6 +218,8 @@
     path("interop-query/query-by-strain", interop_views.query_by_strain, name="query_by_strain"),
     path("interop-query/query-by-gene", interop_views.query_by_gene, name="query_by_gene"),
     path("interop-query/query-by-pair", interop_views.query_by_pair, name="query_by_pair"),
+    path("interop-query/genes", interop_views.genes, name="genes"),
+    path("interop-query/strains", interop_views.strains, name="strains"),
 
     # path('admin/', admin.site.urls)   # make the amdin panel inaccessible via its utl (the admin admin is preserved for the potential future needs)
 ]

gene_function/models.py

@@ -127,6 +127,21 @@ def get_gene_info_and_pangenomic_class(genome_match, projection=None):
 class GenomeInfo:
     objects = database.MongoDBObjects("pankb_genome_info")
 
+    @staticmethod
+    def get_all_strains():
+        """
+        Return a sorted list of distinct genome_id values.
+        """
+        pipeline = [
+            {"$group": {"_id": "$genome_id"}},
+            {"$sort": {"_id": 1}},
+            {"$project": {"_id": 0, "genome_id": "$_id"}},
+        ]
+
+        cursor = GenomeInfo.objects.aggregate(pipeline)
+        return [doc["genome_id"] for doc in cursor if doc.get("genome_id")]
+
+
     def get_genome_and_isolation_info_pipeline(genome_match):
         return [
             {"$match": genome_match},

interop_query/views.py

@@ -10,6 +10,29 @@
 
 logger = logging.getLogger(__name__)
 
+@csrf_exempt
+@require_http_methods(["GET"])
+def genes(request):
+    """Return a flat list of all gene names."""
+    try:
+        genes = GeneAnnotations.get_all_genes()
+        return JsonResponse(genes, safe=False)
+    except Exception as e:
+        logger.exception("list_all_genes failed")
+        return JsonResponse({"message": f"Error: {e}"}, status=500)
+
+
+@csrf_exempt
+@require_http_methods(["GET"])
+def strains(request):
+    """Return all strains (genomes) with isolation info merged."""
+    try:
+        data = GenomeInfo.get_all_strains()
+        return JsonResponse(data, safe=False)
+    except Exception as e:
+        logger.exception("list_all_strains failed")
+        return JsonResponse({"message": f"Error: {e}"}, status=500)
+
 
 @csrf_exempt
 @require_http_methods(["POST"])

pangenome_analyses/models.py

@@ -4,6 +4,18 @@
 class GeneAnnotations:
     objects = database.MongoDBObjects('pankb_gene_annotations')
 
+    @staticmethod
+    def get_all_genes(projection=None):
+        """
+        Return a sorted list of distinct gene names.
+        """
+        cursor = GeneAnnotations.objects.find(
+            {}, projection=projection or ["gene"]
+        )
+
+        names = {doc.get("gene") for doc in cursor if "gene" in doc}
+        return sorted(n for n in names if n)
+
     def get_gene_analysis_pairs(genes):
         """
         Return all distinct (gene, pangenome_analysis) tuples