Merge pull request #28 from TieuLongPhan/dev

Dev

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "synrbl"
-version = "0.0.15"
+version = "0.0.16"
 authors = [
 	{name="Tieu Long Phan", email="long.tieu_phan@uni-leipzig.de"}, 
 	{name="Klaus Weinbauer", email="klaus@bioinf.uni-leipzig.de"}

synrbl/SynCmd/cmd_benchmark.py

@@ -123,6 +123,7 @@ def run(args):
 
     rb_correct = 0
     mcs_correct = 0
+    mcs_cth = 0
     for i, entry in enumerate(dataset):
         if not entry["solved"]:
             continue
@@ -134,6 +135,8 @@ def run(args):
             continue
         exp_reaction = normalize_smiles(exp)
         act_reaction = normalize_smiles(entry[args.col])
+        if entry["confidence"] >= args.min_confidence:
+            mcs_cth += 1
         if (
             wc_similarity(exp_reaction, act_reaction, args.similarity_method)
             >= args.similarity_threshold
@@ -143,6 +146,7 @@ def run(args):
             elif entry["solved_by"] == "mcs-based":
                 mcs_correct += 1
 
+    stats["confident_cnt"] = mcs_cth
     output_result(stats, rb_correct, mcs_correct, file=args.o)
 
 

synrbl/balancing.py

@@ -88,7 +88,7 @@ def __init__(
             mcs_data_col=self.__mcs_data_col,
             issue_col=self.__issue_col,
             rules_col=self.__rules_col,
-            smiles_standardizer=[MoleculeStandardizer()]
+            smiles_standardizer=[MoleculeStandardizer()],
         )
         self.post_processor = PostProcess(
             id_col=id_col, reaction_col=reaction_col, n_jobs=n_jobs, verbose=0
@@ -114,7 +114,10 @@ def __post_process(self, reactions):
         ]
         pp_results = self.post_processor.fit(pp_data)
         for pp_result in pp_results:
-            if pp_result["label"] != "unspecified":
+            if (
+                pp_result["label"] != "unspecified"
+                and "curated_reaction" in pp_result.keys()
+            ):
                 idx = key_index_map[pp_result[self.__id_col]]
                 reactions[idx][self.__reaction_col] = pp_result["curated_reaction"]