Announcing a new dataset by OpenFold Principal Investigator, @grocklin! The Rocklin Lab has released the MGnify Stability Dataset: folding stability measurements for 1.8 million diverse protein domains spanning more than 200,000 sequence families.
We @open_fold are excited for the first major data release from @grocklin partly funded by our consortium. First but by no means the last! Large scale, high quality, diverse stability datasets like this are crucial for more useful protein #ai for #biotech https://t.co/0m7xUGojSl
In keeping with our commitment to open science, all resulting code will be released under permissive licenses so researchers and companies worldwide can build on it. Read more in our linked press announcement! https://t.co/VbTDD7hWte
In collaboration with the Institute for Protein Design, University of Washington led by Prof. David Baker, we are launching a new research fellowship to advance open-source AI for protein structure prediction and design! More 👇
The fellowship will support graduate students and postdoctoral scholars in the Baker Lab working on next-generation models for antibody-antigen design and structure prediction, with OpenFold engineers helping package, document, and maintain the resulting software.
First, the numbers. On small molecule performance assessed by Runs N’ Poses, OF3p2 is comparable to AF3 and the recent Protenix-v1 (PX) from ByteDance, the best other AF3 repro.
Boltz-2 uses more training data and is thus advantaged but performs similarly. 2/9
New OpenFold3 preview out! (OF3p2)
It closes the gap to AlphaFold3 for most modalities.
Most critically, we're releasing everything, including training sets & configs, making OF3p2 the only current AF3-based model that is functionally trainable & reproducible from scratch🧵1/9