()
| 544 | return graph, dpi_dict, cpi_dict |
| 545 | |
| 546 | def process_smiles(): |
| 547 | #padding for transynergy/transformer |
| 548 | smiles_list = [] |
| 549 | for mol in molecules: |
| 550 | smiles = mol.GetProp('SMILES') |
| 551 | smiles_list.append(smiles) |
| 552 | #Convert SMILES's character into index |
| 553 | seqs=smiles_list |
| 554 | chars=set([char for seq in seqs for char in seq]) |
| 555 | chars = list(chars) #for zero embedding |
| 556 | |
| 557 | smiles_dict = dict() |
| 558 | for mol in molecules: |
| 559 | drugbank_id = mol.GetProp('DATABASE_ID') |
| 560 | smiles = mol.GetProp('SMILES') |
| 561 | |
| 562 | d1_sm = [Mapping(chars).item2idx[char] for char in smiles] |
| 563 | smiles_dict[split_it(drugbank_id)] = d1_sm |
| 564 | #len(Mapping(chars).idx2item) |
| 565 | return smiles_dict |
| 566 | |
| 567 | |
| 568 | def process_smiles2graph(): |
no test coverage detected