Search-Based Generation

Search-based and evolutionary methods for molecular generation, including genetic algorithms operating on SMILES strings or molecular graphs and training-free mutation strategies on SELFIES.

Year	Paper	Key Idea
2018	ChemGE	Grammatical evolution over SMILES CFGs for parallel molecule generation
2019	Graph-Based GA and MCTS	Graph-based genetic algorithm outperforming ML methods for penalized logP
2021	STONED	Training-free SELFIES string mutations for molecular design
2023	GAs as Baselines for Molecule Generation	The GA criterion: new methods should beat genetic algorithm baselines

Computational Chemistry

ChemGE pipeline from integer chromosome through CFG grammar rules to valid SMILES output

ChemGE: Molecule Generation via Grammatical Evolution

ChemGE uses grammatical evolution over SMILES context-free grammars to generate diverse drug-like molecules in parallel, outperforming deep learning baselines in throughput and molecular diversity.

Computational Chemistry

Comparison bar chart showing penalized logP scores for GB-GA, GB-GM-MCTS, and ML-based molecular optimization methods

Graph-Based GA and MCTS Generative Model for Molecules

A graph-based genetic algorithm (GB-GA) and a graph-based generative model with Monte Carlo tree search (GB-GM-MCTS) for molecular optimization that match or outperform ML-based generative approaches while being orders of magnitude faster.

Computational Chemistry

Visualization of STONED algorithm generating a local chemical subspace around a seed molecule through SELFIES string mutations, with a chemical path shown between two endpoints

STONED: Training-Free Molecular Design with SELFIES

STONED introduces simple string manipulation algorithms on SELFIES for molecular design, achieving competitive results with deep generative models while requiring no training data or GPU resources.

Computational Chemistry

Diagram showing a genetic algorithm for molecules where a parent albuterol molecule undergoes mutation to produce two child molecules, with a selection and repeat loop

Genetic Algorithms as Baselines for Molecule Generation

This position paper demonstrates that genetic algorithms (GAs) perform surprisingly well on molecular generation benchmarks, often outperforming complex deep learning methods. The authors propose the GA criterion: new molecule generation algorithms should demonstrate a clear advantage over GAs.