Inability to process SMILES with certain SMILES characters

[sohaibbu2015]

Hello, My team and I are interested in using your package to facilitate the generation of new molecules with potential to have a certain type of toxicity. In order to do that, we explored the ability of your package to use a user-defined scoring function as the center of the training protocol. In addition to that, a new dataset is required to complete the training process. Although your package proved to be able to carry out training using user-defined functions, it seems to have some issues handling SMILES representations that contain certain characters. After some investigation, it seems like the package fails to process SMILES that contain / and \ characters which are used to indicate the cis and trans positions of atoms. we were wondering if there exist an easy fix to this problem and if yes, what should be done to fix that issue.

This the error it keeps showing data processing 0/44 Traceback (most recent call last): File "main.py", line 235, in learn(mol_sml, args) File "main.py", line 121, in learn subgraph_set_init, input_graphs_dict_init = data_processing(smiles_list, args.GNN_model_path, args.motif) File "/home/qspt_user/data_efficient_grammar/grammar_generation.py", line 42, in data_processing subgraphs.append(SubGraph(subgraph_i_mapped, mapping_to_input_mol=subgraph_i_mapped, subfrags=list(cluster))) File "/home/qspt_user/data_efficient_grammar/private/molecule_graph.py", line 91, in init super(SubGraph, self).init(mol, is_subgraph=True, mapping_to_input_mol=mapping_to_input_mol) File "/home/qspt_user/data_efficient_grammar/private/molecule_graph.py", line 15, in init self.hypergraph = mol_to_hg(mol, kekulize=True, add_Hs=False) File "/home/qspt_user/data_efficient_grammar/private/hypergraph.py", line 744, in mol_to_hg bipartite_g = mol_to_bipartite(mol, kekulize) File "/home/qspt_user/data_efficient_grammar/private/hypergraph.py", line 692, in mol_to_bipartite mol = standardize_stereo(mol) File "/home/qspt_user/data_efficient_grammar/private/hypergraph.py", line 938, in standardize_stereo atom_idx_1 = each_bond.GetStereoAtoms()[0] IndexError: Index out of range

[gmh14]

Hey,

My current suggestion is to remove these "/", "", or "*" from the SMILES strings since we now cannot handle any stereo/geometry information of molecular graphs.