Amino acid SMILES to be represented as zwitterions

[jonwzheng]

In the current version of the dataset, amino acids are represented in their neutral molecular form. But in real-world applications, amino acids are far more stable in their zwitterionic form in solution.

Future work should be systematically modify those SMILES strings to be zwitterionic (this shouldn't be a too big challenge, and can definitely be done i.e. using RDKit, substructure matching for the amino acid group and modifying those atoms. Or even more naively, modifying the SMILES strings.)