When Molecular GAN Meets Byte-Pair Encoding
This addresses the challenge of generating novel molecular structures for drug discovery, though it appears incremental as it builds on existing GAN and tokenization methods.
The paper tackled the problem of generating novel drug-like molecules by developing a molecular GAN that uses byte-pair encoding tokenization and reinforcement learning, resulting in improved performance in validity, uniqueness, novelty, and diversity metrics.
Deep generative models, such as generative adversarial networks (GANs), are pivotal in discovering novel drug-like candidates via de novo molecular generation. However, traditional character-wise tokenizers often struggle with identifying novel and complex sub-structures in molecular data. In contrast, alternative tokenization methods have demonstrated superior performance. This study introduces a molecular GAN that integrates a byte level byte-pair encoding tokenizer and employs reinforcement learning to enhance de novo molecular generation. Specifically, the generator functions as an actor, producing SMILES strings, while the discriminator acts as a critic, evaluating their quality. Our molecular GAN also integrates innovative reward mechanisms aimed at improving computational efficiency. Experimental results assessing validity, uniqueness, novelty, and diversity, complemented by detailed visualization analysis, robustly demonstrate the effectiveness of our GAN.