Randomizing molecule building blocks improves small dataset molecule creation
What to randomize matters: a controlled study of volume, axis and representation in low-data molecular generation
Machine LearningComputational Engineering, Finance, and ScienceComputation and Language
Summary
Making new molecules with computers is hard when starting with very few examples, because the system tends to copy what it already knows. The authors studied ways to mix up how molecules are represented by changing the order of atoms and larger pieces during training. They found that randomizing both these parts together helps the system create many more truly new molecules than just randomizing one part or using the usual methods. This means what you choose to randomize is more important than how much you randomize for making new molecules from small data.
What this means in practice
- •For computational chemists: Use randomizing atom and fragment order in fragment-based molecular models to greatly increase novel molecule discovery with limited experimental data.
- •For pharmaceutical data teams: Implement fragment-order randomization in drug candidate generation workflows to improve the variety of molecules proposed from small compound libraries.
Authors
Neerav Kaushal, Alisa Teige, Ajay MNV Penmatsa
Abstract
Chemical language models can struggle to generate new molecules when trained on small datasets, where a model may reproduce training examples rather than learn a useful generative distribution. Randomized SMILES augmentation varies atom traversal, whereas Sequential Attachment-based Fragment Embedding (SAFE) also allows the order of fragment blocks to vary. How these distinct sources of representational variation affect low-data molecular generation has not been clearly studied. Here, using 1,000 seed molecules to train an autoregressive sequence model, we compare canonical SAFE with randomized SAFE at increasing corpus sizes, isolate atom-order and fragment-order axes of SAFE randomization, and compare fully randomized SAFE with randomized SMILES at molecule-by-molecule matched training corpora. Three findings emerge. First, randomization of SAFE alone raises the genuine yield of novel molecules more than 450-fold, turning a model that merely reproduces its training set into a practical molecular generator. Second, the two axes of randomization act jointly rather than additively: atom order alone reaches a tenth, and fragment order alone a seventh, of the genuine yield obtained when both vary, which is four times their combined contribution. Third, at matched corpus size and compute, randomized SAFE delivers 2.8 times the genuine yield of randomized SMILES. Welch tests with Games-Howell post-hoc comparisons resolve every difference on the primary outcomes. In the low-data regime, thus, what one randomizes rather than only how much determines whether augmentation produces new molecules or merely new spellings of the training set.