A Genetic Algorithm for Navigating Synthesizable Molecular Spaces
Alston Lo, Connor W. Coley, Wojciech Matusik · 2025
Read A Genetic Algorithm for Navigating Synthesizable Molecular Spaces by Alston Lo, Connor W. Coley, Wojciech Matusik (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-20719.txt.
Description
A molecule is only useful in drug work if chemists can actually make it. Alston Lo, Connor Coley, and Wojciech Matusik present SynGA in a 2025 paper written for ICLR 2026. SynGA is a genetic algorithm that searches directly over synthesis routes rather than raw structures.
Custom crossover and mutation steps keep every candidate inside synthesizable space by construction. Changing the fitness function adapts the search to different jobs. The authors test it on finding synthesizable analogs of known compounds and on optimizing properties with few evaluations. A machine learning filter that narrows the building block set lifts performance further. A model-based variant pairs the search with Bayesian optimization for property tasks.
Catalog record
| CREATOR | Alston Lo, Connor W. Coley, Wojciech Matusik |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 69 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-20719