Reposted in edited form b/c I realized my original post could have been misinterpreted.

For me, it was the referee who used AI. Though AI generated papers and edits are certainly a problem.

I just realized my original post could be read the wrong way. Paper was from my lab and out for review in a high profile journal. Four years of work by a dozen people in multiple labs and AI slop seem to have been the judge.

This was a great collaboration with the lab of Virginijus Siksnys!

This was an incredible collaboration with the labs of Randall Platt, Gerald Schwank, Steve Jackson, Helmuth Gehart, and Timm Schroeder. Special thanks also to the NOMIS Foundation who supported some of the very ambitious experiments.

Definitely true that epigenetic state can affect repair. We try to control for this by editing in highly expressed housekeeping genes with approx equal epigenetic states in all cells. And HDR is just one change. There are stereotyped differences in categorical and molecular end joining outcomes.

Generally true, but I'm pretty sure we're not seeing sampling error. If that were true, then equally populated cell subtypes would have similar differential outcomes. Instead, we can recover cell type solely by clustering on differential outcomes! And "stemmy" cells share outcomes across lineages.

There's some crazy stuff in this preprint. For example, did you know that some neuronal subsets actually do HDR with surprising efficiency?

I’m still learning Bluesky, so now tagging people I mis-tagged. Sorry! @charlesyeh.bsky.social @albertomarin.bsky.social @fedeteloni.bsky.social @gerlichlab.bsky.social

If you’re interested in how loop-extruding cohesin drives the homology search during HR, check out also this new preprint from albertomarin.bsky.social in the Taekjip Ha lab (doi.org/10.1101/2025...)

Curious about how cohesin guides homology search via loops & sister chromatid linkages? Check out this new preprint by fedeteloni.bsky.social & gerlichlab.bsky.social! 👉 tinyurl.com/DSBcohesin