cleaning, fixing, linting

Author: florian-huber

ms2query/database/ann_vector_index.py

@@ -727,7 +727,7 @@ def save_index(self, path_prefix: str) -> None:
         if self._index is None:
             raise RuntimeError("Index not built.")
 
-        # Also save data so that loadIndex(..., load_data=True) works
+        # Also save data so that load_index(..., load_data=True) works
         self._index.saveIndex(f"{path_prefix}.nmslib", save_data=True)
         np.save(f"{path_prefix}.ids.npy", self._comp_ids)
 

ms2query/database/compound_database.py

@@ -467,6 +467,15 @@ def compute_fingerprints(
         Does NOT write to the database.
         Provide exactly one of (smiles, inchis).
 
+        Parameters
+        ----------
+        smiles : Optional[List[str]], optional
+            List of SMILES strings, by default None
+        inchis : Optional[List[str]], optional
+            List of InChI strings, by default None
+        progress_bar : bool, optional
+            Whether to show a progress bar, by default False
+
         Returns the same shapes/types as compute_morgan_fingerprints:
           - dense: np.ndarray of shape (N, nbits)
           - sparse/binary: List[np.ndarray[uint32]]

ms2query/ms2query_library.py

@@ -185,25 +185,29 @@ def query_spectra_by_spectra(
 
     def query_compounds_by_compounds(
         self,
-        compounds: Sequence[str],
         *,
+        smiles: Optional[List[str]] = None,
+        inchis: Optional[List[str]] = None,
         k_compounds: int = 10,
     ) -> List[List[Dict[str, Any]]]:
         """
         Query the fingerprint index with compounds, return top-k compounds per compound.
 
         Parameters
         ----------
-        compounds : Sequence[str]
-            Query compounds (expects list of SMILES strings).
+        smiles : Optional[List[str]], optional
+            List of SMILES strings, by default None
+        inchis : Optional[List[str]], optional
+            List of InChI strings, by default None
         k_compounds : int
             Number of top compounds to return per query compound.
         """
         self._ensure_fingerprint_index()
 
         # Compute fingerprints (sparse representation)
         fps = self.db.all_cdb.compute_fingerprints(
-            compounds,
+            smiles=smiles,
+            inchis=inchis,
         )
 
         # Batched fingerprint ANN query