Thread by @ideasbyjin

Pretty wild that there have been at least 4-5 antibody-specific structure prediction tools this year, all based on #deeplearning

1. DeepAb/IgFold (@jeffruffolo)
2. ABLooper/ImmuneBuilder (@brennanaba)
3. Equifold (@jaehyeon_lee_ml)
4. t-AbFold

What does this mean? 🧵(1/6)

First, it's pretty crazy we even have antibody-specific tools, since #AlphaFold2, #ESMFold, #OmegaFold, all do a decent job at antibody modelling. However, antibody-specific tools have -some- feature that's necessary (e.g. being MSA-free) (2/6)

The demand is likely due to interest from pharma & biotech, but we don't have anywhere near the same level of interest for other polymorphic proteins like TCRs and MHCs (🤔). Regardless, with such interest, I think an antibody-specific CASP should be resurrected! (3/6)

There were the antibody modelling assessment (AMA) competitions, i.e. an "antibody CASP", which was last held ~6 years ago?? The benefit of a competition is having a more objective, held out test set, that makes it easier to compare these antibody specific tools (4/6).

BUT, why don't we go further still? Instead of just saying "whoop I got a lower RMSD than you in competition X", why don't we conjure an assessment that uses structural models to propose mutations for antibodies, or test if we can capture alternate conformations, etc? (5/6)

In sum, the huge interest in antibody structural modelling approaches, I think, calls for a competition. We shouldn't just look at computational metrics, but also experimental ones that will ultimately help biologists make more informed decisions to design the best drugs (6/6)

Correction: it was tFold-Ab, and I missed xTrimoAbFold.