G3-421: Add mapping fix for symbol differences between AON and GW

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "geneweaver-client"
-version = "0.10.0"
+version = "0.10.1a0"
 description = "A Python Client for the Geneweaver API"
 authors = ["Jax Computational Sciences <[email protected]>"]
 readme = "README.md"

src/geneweaver/client/api/mapping.py

@@ -1,10 +1,12 @@
 """Cross-API Geneset Symbol Mapping."""
 
+import re
 from typing import List, Optional
 
 from geneweaver.client.api import aon, genes, genesets
 from geneweaver.client.utils.aon import map_symbols
 from geneweaver.core.enum import GeneIdentifier, Species
+from geneweaver.core.mapping import AON_ID_TYPE_FOR_SPECIES
 
 
 def ensembl_mouse_mapping(
@@ -27,11 +29,10 @@ def ensembl_mouse_mapping(
 
     gene_id_type = GeneIdentifier.ENSEMBLE_GENE
 
-    if species == Species.HOMO_SAPIENS:
-        gene_id_type = GeneIdentifier.HGNC
+    if species != Species.MUS_MUSCULUS:
+        gene_id_type = AON_ID_TYPE_FOR_SPECIES[species]
 
     response = genesets.get_values(access_token, geneset_id, gene_id_type, in_threshold)
-
     if species == Species.MUS_MUSCULUS:
         result = [
             {"gene_id": item["symbol"], "score": item["value"]}
@@ -44,6 +45,7 @@ def ensembl_mouse_mapping(
         else:
             algorithm_id = None
 
+        response = clean_identifiers_for_aon(response, species)
         aon_response = aon.ortholog_mapping(
             [g["symbol"] for g in response["data"]],
             Species.MUS_MUSCULUS,
@@ -73,3 +75,57 @@ def ensembl_mouse_mapping(
         result = [{"gene_id": k, "score": v} for k, v in ensembl_result.items()]
 
     return result
+
+
+IDENTIFIER_PREFIX_MAP = {
+    Species.DANIO_RERIO: "ZFIN",
+    Species.DROSOPHILA_MELANOGASTER: "FB",
+    Species.CAENORHABDITIS_ELEGANS: "WB",
+    Species.SACCHAROMYCES_CEREVISIAE: "SGD",
+}
+
+
+def clean_identifiers_for_aon(data: dict, species: Species) -> dict:
+    """Clean up identifiers for AON mapping."""
+    if species in [
+        Species.DANIO_RERIO,
+        Species.DROSOPHILA_MELANOGASTER,
+        Species.CAENORHABDITIS_ELEGANS,
+        Species.SACCHAROMYCES_CEREVISIAE,
+    ]:
+        data = {
+            "data": [
+                {
+                    "symbol": f"{IDENTIFIER_PREFIX_MAP[species]}:{item['symbol']}",
+                    "value": item["value"],
+                }
+                for item in data["data"]
+            ]
+        }
+    elif species == Species.RATTUS_NORVEGICUS:
+        data = {
+            "data": [
+                {
+                    "symbol": insert_colon_delimiter(item["symbol"]),
+                    "value": item["value"],
+                }
+                for item in data["data"]
+            ]
+        }
+
+    return data
+
+
+def insert_colon_delimiter(identifier: str) -> str:
+    """Separates a prefix from ID and adds a colon delimiter between them."""
+    if ":" in identifier:
+        return identifier
+
+    match = re.match(r"([A-Za-z]+)(\d+)", identifier)
+
+    if match:
+        prefix, suffix = match.groups()
+        return f"{prefix}:{suffix}"
+    else:
+        # If the identifier does not match the expected pattern
+        raise ValueError("Identifier format is invalid")