3 ms·
AlphaFold predictions: great hypotheses but no match for experiment
- Rochus 4y agoInteresting findings; so the NMR structure researchers won't be out of a job after all; good to know.
- dekhn 4y agoThe paper doesn't mention NMR at all, and in fact, AlphaFold filtered out all NMR structures before training. NMR protein structures are mostly useless in terms of large-scale structure but they are useful for probing dynamics and fine details of some aspects of smaller proteins.
- Rochus 4y agoRight; but if crystallograhers keep their jobs, also Bio NMR folks will likely be lucky; NMR can well be used to validate proposed structures even without being forced to first create crystals.
- dekhn 4y agoUnless something important has changed, I don't think NMR is very useful for protein structure above about 30-40kDa and even then typically requires heroic technique to obtain more than a few details about torsions and interatomic distances.
- Rochus 4y agoI was merely expressing my satisfaction that experimental structure determination is still needed. Thus, my work continues to have some value. And there was indeed some progress in the last 25 years.