Merge pull request #1429 from bgyori/mgi_ensembl

Add Ensembl info for MGI genes

indra/databases/mgi_client.py

@@ -93,13 +93,32 @@ def get_id_from_name_synonym(name_synonym: str) -> Union[None, str, List[str]]:
     return None
 
 
+def get_ensembl_id(mgi_id: str) -> Union[str, None]:
+    """Return the Ensembl ID for an MGI ID.
+
+    Parameters
+    ----------
+    mgi_id :
+        An MGI ID, without prefix.
+
+    Returns
+    -------
+    :
+        The Ensembl ID corresponding to the MGI ID, or None if not available.
+    """
+    if mgi_id and mgi_id.startswith('MGI:'):
+        mgi_id = mgi_id[4:]
+    return mgi_id_to_ensembl.get(mgi_id)
+
+
 def _read_mgi():
     fname = get_resource_path('mgi_entries.tsv')
     mgi_id_to_name = {}
     mgi_name_to_id = {}
     mgi_synonyms = {}
     mgi_synonyms_reverse = defaultdict(list)
-    for mgi_id, name, synonyms_str in read_unicode_csv(fname, '\t'):
+    mgi_id_to_ensembl = {}
+    for mgi_id, name, synonyms_str, ensembl_id in read_unicode_csv(fname, '\t'):
         if name:
             mgi_id_to_name[mgi_id] = name
             mgi_name_to_id[name] = mgi_id
@@ -108,9 +127,12 @@ def _read_mgi():
             mgi_synonyms[mgi_id] = synonyms
             for synonym in synonyms:
                 mgi_synonyms_reverse[synonym].append(mgi_id)
+        if ensembl_id:
+            mgi_id_to_ensembl[mgi_id] = ensembl_id
 
     return mgi_id_to_name, mgi_name_to_id, mgi_synonyms, \
-        dict(mgi_synonyms_reverse)
+        dict(mgi_synonyms_reverse), mgi_id_to_ensembl
 
 
-mgi_id_to_name, mgi_name_to_id, mgi_synonyms, mgi_synonyms_reverse = _read_mgi()
+mgi_id_to_name, mgi_name_to_id, mgi_synonyms, mgi_synonyms_reverse, \
+   mgi_id_to_ensembl = _read_mgi()