Hey everyone! We will be having an Office Hour next week on 6/17/2025 at 1 PM EDT!(https://www.timeanddate.com/worldclock/fixedtime.html?msg=Foldit+office+hour&iso=20250617T13&p1=43&ah=1).
We will be joined by the UC Davis researchers from the recent Unsolved Voltage-gated Ion Channel Cryo-EM Density puzzles. You can read their blogpost from last month here: https://fold.it/forum/blog/foldit-players-help-model-a-spider-toxin-for-pain-therapy
Office Hour Transcript, thanks to LociOiling:
https://foldit.fandom.com/wiki/17June_2025-_Office_hour_with_Dr._Vladimir_Yarov-Yarovoy,_dlopezma,_and_beta_helix
Let's try that link again:
17 June 2025 - Office hour with Dr. Vladimir Yarov-Yarovoy, dlopezma, and beta helix
I got stuck at work and couldn't make this one. Hopefully i can make the next one.
It's always nice to ask what general idea do you think the structures of Fold It need to improve at? Is there any part of the fold that seemed off or missing?
Thanks for posting the Transcript. I found the Lipinski's Rules as an interesting concept to look into. seems like something a recipe could calculate with some clever maths.
No worries about missing the session—life happens! I’m glad the transcript (and Lipinski’s Rules) caught your eye; those “rules of five” really are a cornerstone in early drug-likeness filtering, and a recipe‐style script can absolutely evaluate them for drug designs with a bit of clever math.
On your question about where Foldit models could improve: because our focus is voltage-gated ion channels—membrane proteins—the biggest gap is that Foldit scores everything as if it were in bulk water. The lipid bilayer’s hydrophobic environment isn’t represented. As a result, transmembrane helices may adopt unrealistic tilts, or charged or polar side-chains aren’t scored as they would be inside the hydrophobic bilayer.
If Foldit had an implicit membrane term—penalizing charged residues buried in the hydrophobic core for example—we’d get models with better quality!