Mathis, S. | Search bioRxiv

bioRxiv2025

Accelerating Biomolecular Modeling with AtomWorks andRF3

Deep learning methods trained on protein structure databases have revolutionized biomolecular structure prediction, but developing and training new models remains a considerable challenge. To facilitate the development of new models, we present AtomWorks: a broadly applicable data framework for developing state-of-the-art biomolecular foundation models spanning diverse tasks, including structure prediction, generative protein design, and fixed backbone sequence design. We use AtomWorks to train RosettaFold-3 (RF3), a structure prediction network capable of predicting arbitrary biomolecular complexes with an improved treatment of chirality that narrows the performance gap between closed-source AlphaFold3 (AF3) and existing open-source implementations. We expect that AtomWorks will accelerate the next generation of open-source biomolecular machine learning models and that RF3 will be broadly useful as a structure prediction tool. To this end, we release the AtomWorks framework (https://github.com/RosettaCommons/atomworks), together with curated training data, code and model weights for RF3 (https://github.com/RosettaCommons/modelforge) under a permissive BSD license.

biochemistry↗

bioRxiv2024

Microdroplet screening rapidly profiles a biocatalyst to enable its AI-assisted engineering

Engineering enzymes for increased efficiency is key to enabling sustainable, green biocatalytic production processes in the chemical and pharmaceutical industries. This challenge can be tackled from two angles: by directed evolution, based on labour-intensive experimental testing of enzyme variant libraries, or by computational methods, where data-dependent algorithms relating sequence and function are used to predict biocatalyst improvements. Here, we combine both approaches into a two-week, low-cost workflow, in which ultra-high throughput screening of a library of imine reductases (IREDs) in microfluidic devices provides not only selected hits, but also long-read sequence data linked to fitness scores of >17 thousand enzyme variants. We demonstrate the engineering of an IRED for chiral amine synthesis by mapping its local fitness landscape in one go, ready to be used for interpretation and extrapolation by protein engineers with the help of machine learning (ML). We calculate position-dependent mutability and combinability scores of mutations and comprehensively illuminate a complex interplay of mutations driven by synergistic, often positively epistatic effects. When interpreted by easy-to-use regression and tree-based ML algorithms designed for random whole-gene mutagenesis data, 3-fold improved hits initially obtained from experimental screening are extrapolated further to give another order of magnitude improvement (23-fold in kcat) after testing only a handful of designed mutants. Predictions succeed in >80% of cases. The catalytic features discovered in one IRED are shown to be portable and confer activity on IREDs with [~]50% homology. Our campaigns yield biocatalytically efficient IREDs and are paradigmatic for future enzyme engineering efforts that rely on large sequence-function maps, profiling how a biocatalyst responds to mutation. In the age of predictive biology, these maps will chart the way to improved function by exploiting the synergy of rapid experimental screening combined with ML evaluation and extrapolation.

synthetic biology↗