linting

ms2deepscore/data_generators.py

@@ -596,13 +596,11 @@ def __len__(self):
             * int(np.floor(len(self.selected_compound_pairs.scores) / self.settings["batch_size"]))
 
     def _spectrum_pair_generator(self, batch_index: int) -> Iterator[SpectrumPair]:
-
-        same_prob_bins = self.settings["same_prob_bins"]
+        """Use the provided SelectedCompoundPairs object to pick pairs."""
         batch_size = self.settings["batch_size"]
-
         indexes = self.indexes[batch_index * batch_size:(batch_index + 1) * batch_size]
         for index in indexes:
-            inchikey1 = self.selected_compound_pairs._idx_to_inchikey[index]
+            inchikey1 = self.selected_compound_pairs.idx_to_inchikey[index]
             score, inchikey2 = self.selected_compound_pairs.next_pair_for_inchikey(inchikey1)
             spectrum1 = self._get_spectrum_with_inchikey(inchikey1)
             spectrum2 = self._get_spectrum_with_inchikey(inchikey2)

ms2deepscore/spectrum_pair_selection.py

@@ -70,6 +70,14 @@ def generator(self):
     def scores(self):
         return self._scores
 
+    @property
+    def idx_to_inchikey(self):
+        return self._idx_to_inchikey
+
+    @property
+    def inchikey_to_idx(self):
+        return self._inchikey_to_idx
+
     def __str__(self):
         return f"SelectedCompoundPairs with {len(self._scores)} columns."
 
@@ -230,8 +238,10 @@ def compute_jaccard_similarity_matrix_cherrypicking(
             scores_data.extend(scores_row[idx_selected])
             scores_i.extend(len(idx_selected) * [i])
             scores_j.extend(list(idx_selected))
+    # Output biases in scores
+    print("Counted cases of fewer than max_pairs_per_bin usage:")
+    max_pairs_global = [x - max_pairs_per_bin for x in max_pairs_global]
     print(max_pairs_global)
-
     return scores_data, scores_i, scores_j