Post

Carl Zimmer
@carlzimmer.com

Here's a new kind of virus that scientists have generated with AI. And here's my story about what it means for us to have gotten to this remarkable moment: nyti.ms/4bBhoWB (Gift link) 🧪

73 reposts146 likes

Kitty Kowalski
@kittykowalski.bsky.social

I've seen this movie before. Then they think they can use this virus against their enemies, and then don't realize that viruses don't choose between the good people and the bad people, and then it spreads and wreaks Havoc. We all know how this movie ends

Alex Crits-Christoph
@acritschristoph.bsky.social

see also this work, which helps provide some helpful framing of how novel these phages are: www.biorxiv.org/content/10.6... tl;dr: I believe this work generates only marginally novel phages, which have only a moderate likelihood of working, that decreases the more novel they get.

biorxiv.orgQuantifying evolutionary novelty and design efficiency in generative genome designGenerative genome design models can now produce previously unobserved genome-length sequences, but assessing their capabilities is complicated by limitations in functional prediction. The ability to e...
Darren Martin
@darrenmartin.bsky.social

I really like this deep mutagenesis paper - Besides being cool it emphasizes just how hard it is to design a viable PhiX-like genome that is even just 20 or so nucleotides diverged from any currently known genomes. These small single stranded DNA virus genomes are very informationally dense .....

Darren Martin
@darrenmartin.bsky.social

...I can't access this new PhX paper in Science, but I would be very impressed if (1) some of the viable genomes were >200 nucleotides diverged from any previously known genomes (2) the differences were novel (not recombinants) and (3) included muts flagged as lethal by the deep mutagenesis study.

Alex Crits-Christoph
@acritschristoph.bsky.social

Yeah, that's ok, but a bit of an understatement. For a better view, this figure from the preprint above - it's kind of a bummer that this sort of clear-eyed analysis didn't make it into the actual paper itself! There's also this figure from the original preprint: bsky.app/profile/acri...

Alex Crits-Christoph@acritschristoph.bsky.social

And then where are these new mutations from? They could be from other related phage, just not found in the best hit? That's Fig 4F. Indeed many of the new mutations are found in other close relatives. Especially, I think, for competition winners like #69 and #100 (unless I am reading this wrong!)

Barbarasynbio
@barbarasybio.bsky.social

AI-generated virus sounds scary to me. Then I read it’s a phage. We should be extremely careful not to fall to the temptation to create new pathogens because it’s cool. The Sars-Cov-2 pandemic should be fresh enough in our memories…

James
@hereswhatithink.bsky.social

Humanity cannot resist touching the stove. Sci-fi writers have spent decades highlighting various potential dangers with a view to steering us away from them before it's too late, and in response we've spent decades charging straight towards them like moths to a flame, because danger intrigues us.

Barbarasynbio
@barbarasybio.bsky.social

Interesting viewpoint. Not sure it’s because danger intrigues us. It may be part of the story. I think we are mostly quite narcissistic and look for glory in any way we can…

James
@hereswhatithink.bsky.social

That too, yes, though it only really applies to those at the wheel. They would take the "unique privilege" of being the one to blow up the sun over a life perceived as inconsequential. It's the rest of us enabling and cheering it on who just want to see what happens and if it's really all that bad.

What?!
@r-s-s.bsky.social

"University of Manchester, who was not involved in the study" "a protein chemist at the University of Oxford who was not involved in the new study" Is there some law that I don't know about that requires Brits that were not involved must be cited in scientific articles in the US?