Kresten Lindorff-Larsen

I’m not quite sure how this leads to an estimate of the thermodynamics of the system though

I agree. Which I think is in line with what I was saying. There appears not to be a commonly accepted definition of what De Novo Design means these days though there are maybe some approaches that many agree is not “de novo”.

Thanks. I agree that the confidence score itself isn’t a good estimate of the global free energy.

Ok. Thanks. In the paper mentioned below we designed a thioredoxin using only a backbone structure as template. The resulting sequence is novel yet easily recognisable as a Trx. But we didn’t use MSAs etc. Not sure this is really de novo

bsky.app/profile/lind...

New functions would be a tough requirement given that many (enzymatic) functions probably exist as minor side activities (also I’m generally in favour of talking about “properties” rather than “functions” but that’s a slightly orthogonal point)

I’m not sure what 0% seq identity would mean, but given how new methods are better (than eg BLAST) at identifying sequence similarities I’m also not sure that this is generally the right metric

bsky.app/profile/mart...

I’m not sure there was ever a broadly accepted definition. In the paper below we designed a thioredoxin starting only with the backbone structure and used only Rosetta (ie no evolution). The sequences are easily recognised as thioredoxins. Is that de novo? We didn’t say so

bsky.app/profile/lind...

Does this in your opinion imply that if the structure exists (or is known to exist) in nature, it’s not de novo?

But I was actually thinking beyond these empirical observations and towards a more theoretical view that would link model probabilities to physical free energies akin to this work:

Zero-shot protein stability prediction by inverse folding models: a free energy interpretation
doi.org/10.48550/arX...

And second if it were truly a (free!) energy function then one might even be able to calculate fractions and convert them into relative free energies.

Thanks. Yes I know that paper well. It’s very nice and important. But I’m thinking beyond that. First, that paper shows that sampling is also guided by the MSA and so it’s important to decouple the effect of the MSA and the “intrinsic energy function”

Protein-ligand complexes are all the rage with OpenBind+other efforts launching

3 amazing papers describing 229 protein-ligand structures just dropped in @actacrystd.iucr.org

A tremendous teaching text, but I disagree on the discussed potential for misuse. Read more

fraserlab.com/2025/07/29/E...

There are days where I feel like both

How your email finds me

🚀 Today we're launching the 'Night Science Institute', a non-for-profit organization to lead a cultural shift in science! You may say we're dreamers 😉, but we think we're not the only ones. Perhaps today you will join us to make the Day Science and Night Science parts of the process live as one!

We just bumped into something very preliminary… but very exciting:

AF2.3 and AF3.0 distograms may potentially reproduce MD-like behavior.

Until we do further tests, you can reach our early insights at www.biorxiv.org/content/10.1...

Far i opråb til Ane Halsboe: Hvorfor skal mine børn klassificeres som andenrangsdanskere? Mine børn kom til verden under min udstationering. Det kan nu få betydning for dem, hvis de skulle ende med at få brug for kontanthjælp.

Endnu et eksempel på, hvor absurd regeringens aftale om kontanthjælp og arbejdspligt er. Den rammer i flæng, fordi regeringen ønsker at diskriminere men ikke må. Ministerens forsøg på bortforklaring klinger hult, når hun taler om “ret og pligt”. #dkpol

politiken.dk/debat/debati...

Contrasting photographs of the night-time skylines of Manhattan (left) and Nijmegen (right), with matching genome-wide association plots underneath each.

Not sure who came up with "Manhattan Plot", but in 2014 I coined the alternative term "Nijmegen Plot" (inspired by the Dutch town where I live) to describe underwhelming results from our earliest genome-wide association scans of language/reading traits.

CAPRI Docking

A small reminder to all structural biologists around working on biomolecular complexes: please consider sharing your complexes as targets for CAPRI - AI has not solved all structure prediction problems and there are still challenges! See www.capri-docking.org/contribute/

Tom Lehrer is dead but his music will live on. In the public domain. Whether it’s Poisoning Pigeons in the Park 🕊️, New Math, or The Elements

What a star

Everyone is celebrating him for his (obviously wonderful) songs but I am also celebrating him for releasing all his works into the public domain and insisting he retained no copyrights over any of it.

Instead, I concentrated on paying attention to papers published on my watch to avoid this situation in the future.

I also discouraged over-the-top press releases although I had very little control over that.

12/n

Why I didn't retract this paper when I was Editor-in-Chief at Science (THREAD 🧵)

The "arsenic life" paper finally got retracted by Science after 15 years. RIP As-DNA. #microsky

Story time! When this paper came out in 2010, I was neck-deep trying to grow SAR11 on defined medium. Everyone thought there was some magic "missing" ingredient. Maybe nucleotide monophosphates? 1/6

Presumably the reviewer has a confidence score of 5

Who is Adam

Intrinsic disorder comes in various flavors ! Using CALVADOS simulations of the Tau protein, we reveal that its disordered dynamics can be interpreted at both residue and domain scales, and that addition of phosphorylations impact both

By @sacquin-mo.bsky.social , Chantal Prévost and I

The last step in a long process on “arsenic life”

The 'arsenic life' retraction has triggered quite an (interesting) debate

I fully agree the retraction of flawed work due to technical faults in method/protocol, where mistakes are not picked up by the review/editorial examination

BUT should it take 15 years?? 1/n

www.science.org/content/blog...

I don’t disagree. Was just surprised to see that the quality of the review process would play any role in the argument. It’s of course good that the journal is taking some responsibility in the process.

I wasn’t involved in the crystallography, but this designed thioredoxin was solved by MR and had 8 molecules in the ASU, one of which was substantially different from the 7 others

doi.org/10.1016/j.jm...