It’s easier to improve from 24 than from 78, isn’t?! 🙂

Maybe too complicated due to many chiral centers (9), but in trying to reproduce 9OQ1, I got the Sucralose with the wrong chirality. The pose was close, though.

www.biorxiv.org/content/10.6...
Apparently, introducing noise at inference and masking MSA still works well

Large-scale collaborative assessment of binding free energy calculations for drug discovery using OpenFE Accurately measuring compound binding affinities is key to driving the pharmaceutical development process. Rigorous physics-based in silico approaches, particularly alchemical free energy methods, hav...

chemrxiv.org/engage/chemr...

Nice large-scale benchmark of OpenFE on both public (vs FEP+) and blinded private datasets. As expected, the private sets are notably more challenging. OpenFE still looks on a good trajectory toward competitiveness. I’d love to see FEP+ and other RBFE stacks evaluated in a true blind competition.

Lol

Congrats!

In Germany works like that too. Everything travel through mail. Sometimes different Offices ask for the same original (no print) document to be sent…

www.nature.com/articles/s41...
Run this and go on vacation?

Amazing

yeah, probably performing comparably bad is easier than comparably good.

Hope your wife is moving with you. My best wishes

had same experience. Something wrong with reasoning? In VScode you can install the copilot extension and select among models (Sonnet, GPT or Gemini).

scaring. In the future will be coupled to voice chatbot (Sesame made impressive improvements in the field).

It's like ending up with a list of ranked docked compounds. Your troubles just started.

In other sectors, somebody may have found a working business offer. We just investigated what was on the market for target identification. Most of the companies are struggling to be sustainable. More or less they all end up with genes and corresponding evidences from literature and omics dbs.

I work for a CRO and we realized that it is difficult to ask for a lot of money just to generate hypothesis. Most clients prefer to invest on wet-lab validation.

Yeah, may be for the same reason why code is not maintained once the paper is published. No reward anymore. Maybe at some point somebody will train a LLM that takes a whole software and returns a debugged/optimized version. That something deserving a Nobel.

Converging mechanism of UM171 and KBTBD4 neomorphic cancer mutations - Nature We show that gain-of-function cancer mutations in the KBTBD4 E3 ligase promote neodegradation of substrates via a shape-complementarity-based mechanism, which converges with the mechanism of action of...

Cool! A molecular glue mimicking a cancer mutation enables gain-of-function degradation. www.nature.com/articles/s41...

Do you think Rosetta energy would correlate with dissociation Freq?

First eigenvector is crying