A recent paper described Boltz-1: Democratizing Biomolecular Interaction Modeling https://www.biorxiv.org/content/10.1101/2024.11.19.624167v2
Boltz-1, an open-source deep learning model incorporating innovations in model architecture, speed optimization, and data processing achieving AlphaFold3-level accuracy in predicting the 3D structures of biomolecular complexes. Boltz-1 demonstrates a performance on-par with state-of-the-art commercial models on a range of diverse benchmarks, setting a new benchmark for commercially accessible tools in structural biology.
I tried installing Boltz-1 on Apple Silicon and whilst I did get it running the it was far from straight-forward. Fortunately, the current version of ChimeraX 1.10 will install Boltz-1 on a Mac.
Making Boltz Structure Predictions in ChimeraX https://www.rbvi.ucsf.edu/chimerax/data/boltz-apr2025/boltz_help.html A ChimeraX graphical user interface (menu Tools / Structure Prediction / Boltz) and ChimeraX command (boltz) are provided to make predictions. There is also a YouTube video showing how to make predictions.
Boltz-2 adds a few more complications and these seem to have been addressed in a pull request, but not yet merged. https://github.com/jwohlwend/boltz/pull/231 and https://github.com/fnachon/boltz
There is an extensive and considered review of Boltz-2 provided by Deepmirror https://www.deepmirror.ai/post/boltz-2-real-drug-targets
Boltz-2 represents a powerful and efficient new approach for structure-based drug design. It’s particularly well-suited for targets that are rigid or already well-characterized structurally – in those cases, it can quickly generate accurate protein–ligand complexes and provide a readout of relative binding affinities, which is incredibly useful for designing and prioritizing drug candidates. On the flip side, for proteins that are “shapeshifters” or not well characterized, Boltz-2 alone may not be enough. In such situations, pairing Boltz-2 with downstream refinement methods (like molecular dynamics simulations or flexible docking), or providing it with some structural hints, can help bridge the gap.
