Great work! As github does not have discussion option I thought would ask here: what would be best practical workflow to invent synthesis-worthy binders for particular protein? Use “naive” model with filters or preference aligned one that is currently available or train one?

Great distraction I would enjoy. Will blossom datasets be provided on website, or I have to register to obtain them?

Wow. Impressive work!

Coral, comes in R and as web server: Metz, K. S. et al. Coral: Clear and Customizable Visualization of Human Kinome Data. Cell Syst. 7, 347–350.e341 (2018).

We took a “no cell shall be left behind” approach which wouldn’t be possible without #Cellpose, #TrackMate by @jytinevez.bsky.social , #Fiji and #rstats analysis and visualization with GlmmTMB, emmeans and #SuperPlots and microscopy cores at UCSD. Plasmids available from @addgene.bsky.social

Our biosensor revealed that PKN2 is constitutively active at the plasma membrane—a previously unappreciated hotspot for basal PKN2 activity! Paralleling ROCK signalling. Not every kinase needs triger for activation. Inhibitor first paradigm is worth looking into.

We compared our newly identified peptide to existing ones (from literature and the kinase substrate atlas) and found that ours shows superior sensitivity—perfect for building a PKN biosensor.
rdcu.be/d7o0G

To find the substrate we dephosphorylated cell lysate, heated up, and then rephosphorylated them to reveal the most suitable PKN substrate peptide (KESTREL). This method helped us discover a new, highly sensitive substrate for PKN.