Cheminformatics articles from across Nature Portfolio

Large language models such as GPT-4 have been approaching human-level ability across many expert domains. GPT-4 can accomplish complex tasks in chemistry purely from English instructions, which may transform the future of chemistry.

Retrosynthesis has served as a playground for computer-aided design for many decades. Computer-aided methods are usually predicated on human-expert rules or learning algorithms that extract the rules from literature data. Now, an approach that bridges the gap between these computer-driven methods and the traditional, intuition-driven, ‘chalk board’ retrosynthetic methods is reported.

Informatics approaches play an increasingly important role in accelerating the advances of modern materials science. A recent study reports the development of predictive machine learning models to guide the de novo design of through-space charge transfer polymers with full-colour-tunable emission.

Owing to the diminishing returns of deep learning and the focus on model accuracy, machine learning for chemistry might become an endeavour exclusive to well-funded institutions and industry. Extending the focus to model efficiency and interpretability will make machine learning for chemistry more inclusive and drive methodological progress.