fix predict

src/predict.py

@@ -46,6 +46,8 @@ def compute_drug_embedding(
     os.chdir("MolecularTransformerEmbeddings")
     vector_list = []
     # embed_dict = get_smiles_embeddings(smiles_list)
+    drugs_without_embed = {}
+    drugs_labels = {}
     for drug_id in drugs:
         from_vectordb = vectordb.get("drug", drug_id)
         if len(from_vectordb) > 0:
@@ -55,22 +57,16 @@ def compute_drug_embedding(
             # df = pd.concat([df, pd.DataFrame(embeddings)], ignore_index = True)
             df.loc[len(df)] = embeddings
             continue
+        else:
+            drug_smiles, drug_label = get_smiles_for_drug(drug_id)
+            drugs_without_embed[drug_smiles] = drug_id
+            drugs_labels[drug_id] = drug_label
 
         drug_smiles, drug_label = get_smiles_for_drug(drug_id)
-        log.info(f"⏳💊 Drug {drug_id} not found in VectorDB, computing its embeddings from SMILES {drug_smiles}")
-        embed_dict = get_smiles_embeddings([drug_smiles])
-
-        # with open("../tmp/drug_smiles.txt", "w") as f:
-        #     f.write(drug_smiles)
-        # os.system("python embed.py --data_path=../tmp/drug_smiles.txt")
-        # o = np.load("embeddings/drug_smiles.npz")
-        # files = o.files  # 1 file
-        # gen_embeddings = []
-        # for file in files:
-        #     gen_embeddings.append(o[file])  # 'numpy.ndarray' n length x 512
-        # vectors = np.stack([emb.mean(axis=0) for emb in gen_embeddings])
-        # # In this case we vectorize one by one, so only 1 row in the array
-        # embeddings = vectors[0].tolist()
+        embed_dict = get_smiles_embeddings([drug_id])
+        # log.info(f"⏳💊 Drug {drugs_without_embed.keys()} not found in VectorDB, computing its embeddings from SMILES {drug_smiles}")
+        # embed_dict = get_smiles_embeddings(list(drugs_without_embed.keys()))
+
         # TODO: add label also?
         embeddings: np.array = embed_dict[drug_smiles].tolist()
         vector_list.append({