Gabriel Axel
banner
gabiaxel.bsky.social
Gabriel Axel
@gabiaxel.bsky.social
Computational protein archeologist, plant-based, cat lord
Course by Manolis Kellis: www.youtube.com/playlist?lis...

Seminars about specific cases from Narayan Lab, University of Michigan: www.youtube.com/@narayanlabu... and Boston Protein Design and Modeling Club: www.youtube.com/@bpdmc
MLCB24 - Machine Learning in Computational Biology Fall 2024 - YouTube
Manolis Kellis + Eric Alm
www.youtube.com
June 13, 2025 at 7:48 AM
I suppose this is an inherent issue of any learning-based tool that doesn't actually perform energy calculations. Maybe an additional step for that can be added, similar to how ColabFold has an AMBER relaxation option? It would still be faster than classical docking/MD overall.
June 8, 2025 at 11:41 AM
Marvelous! I tried to use Boltz-2 to compare the binding affinities of different cytokinins to CHASE domain, the results were mostly in line with existing work (doi.org/10.1111/j.13...), except for cis and trans zeatin the predicted affinities were similar while expected to differ significantly.
The specificity of cytokinin signalling in Arabidopsis thaliana is mediated by differing ligand affinities and expression profiles of the receptors
Arabidopsis thaliana has three membrane-located cytokinin receptors (AHK2, AHK3 and CRE1/AHK4), which are sensor histidine kinases containing a ligand-binding CHASE domain. Despite their structural ...
doi.org
June 8, 2025 at 11:41 AM
I experienced a similar exertion. I think an important question to ask is whether they are interested in developing new computational methods (i.e. "the next AlphaFold"), or implementing, modifying and assembling existing (and future) computational methods to answer specific biological questions.
March 2, 2025 at 10:33 PM