Oli Clarke
banner
olibclarke.bsky.social
Oli Clarke
@olibclarke.bsky.social
CryoEM, membrane proteins and whatnot
Is anyone else having problems with NCBI tools? I'm trying to use the CDD database, and it seems completely broken for requests that worked fine before - not sure if shutdown related?
October 21, 2025 at 6:52 PM
I ❤️ NYC
October 18, 2025 at 3:24 PM
Fascinating. And I’m a sucker for beautiful pen and ink diagrams
October 18, 2025 at 12:37 AM
Excellent recommendation, thanks @alisterburt.bsky.social! 🙏 🤓
October 18, 2025 at 12:27 AM
...assess the strength of presented data based on analysis of the figures. E.g. if I was a reviewer for this paper, I would look at this fig and dismiss the paper out of hand, because the maps are nonsense. This is an extreme example, but similar principles apply to analysis of most #structbio data.
October 16, 2025 at 4:00 PM
A tour de force - spectacular work! What is the 70S dome complex, sounds intriguing…? 👀
October 15, 2025 at 11:32 AM
Take a look at the supps… 😬
October 13, 2025 at 1:33 PM
And it's available to borrow/view at archive.org, nice!

archive.org/details/cell...
October 11, 2025 at 3:33 PM
Also this hexagonal thing, and these short bow-ties (kind of reminiscent of TPP2 but shorter?). Also 400Å circle here
October 11, 2025 at 3:20 PM
here with a 400Å circle you can see a little more detail on the sides
October 11, 2025 at 3:18 PM
possibly stupid question - does anyone in #teamtomo or #cryoem recognize these little stacks (from EMPIAR-12549, U2OS cell lamellae)? The green circle is 300Å, and the spacing between the layers is ~50Å. The fourth, bell-shaped thing is something else, also not sure of that one..
October 11, 2025 at 3:10 PM
no wonder - the Fab-antigen interaction in that structure is only resolved at very low resolution, not at the level of sidechain detail. Probably the starting model is to blame...?
October 9, 2025 at 1:33 PM
Turtle yoga in central park 🧘‍♂️
October 5, 2025 at 6:56 PM
Same at HHS…
October 2, 2025 at 10:32 PM
Tried this starting directly from blob-picked particles - no 2D classification. Still works.

The SGD meanders around for 5000 iterations going nowhere, and then suddenly starts to converge at iteration 5400 (this animation ends at iter 5900):
October 2, 2025 at 6:47 PM
some of the ones in the lower left may be peroxiredoxin - we see it as contaminant in recombinant apofe (here easier to visualize with the ice thickness gradient)
September 29, 2025 at 10:40 PM
Beautiful blue winged wasp (Scolia dubia) in central park 🐝
September 29, 2025 at 9:41 PM
“Side chain densities” … 🫣
September 26, 2025 at 2:28 PM
Finding the right params took a bit of tweaking - the batchsize, fraction of initial & final iterations, and the tau2_fudge all need tuning - but once optimized, it converged on a map with clear secondary structure, and local refinement from the final data star gave the map in the previous post.
September 26, 2025 at 2:28 AM
Initial attempt at replicating in relion (parameters in next post). This is for Aca2-RNA, using a 100k subset of the 2D-classified particles (no prior 3D cleanup).

A 1-class ab initio in relion, then local refinement in relion (1.8deg searches+blush) gives a nominally 3.3Å map; 3.5 Å w/out blush.
September 26, 2025 at 2:28 AM
September 25, 2025 at 11:28 PM
Beautiful 🤩 I love that there are still mysteries like this waiting to be unraveled..

(also how much does this look like an impressionist portrait of GroEL-GroES??)
September 23, 2025 at 6:05 PM
seems like adjusting these and dramatically increasing the number of VDAM iterations should be roughly equivalent?
September 21, 2025 at 2:27 PM
Yes although later USCIS guidance says it doesn’t apply to renewals- only new first time H1Bs
September 20, 2025 at 10:56 PM
Replacing green cards with a $1M cash payment… wtf…???
September 20, 2025 at 7:28 PM