James Zou

speaker

371 appearances 1 recordings 1 series first heard Dec 2024 last heard Dec 2024

James Zou’s voice in public audio — every appearance, attributed to the second.

Trend

recordings per month · last 12 months

No recordings in the last 12 months.Older appearances are listed below; set an alert to hear about the next one.

Appearances

newest first · ▶ plays the moment

"The Cognitive Revolution" · Breakthroughs in AI for Biology: AI Lab Groups & Protein Model Interpretability with Prof James Zou · 18 Dec 2024 podcast

So basically what they do is first they would use this protein language model, ESM, to score among like a large list of potential mutations, what tends to be the most promising in terms of likely producing stable and functional proteins.

But those stable functional proteins may not really be useful for the specific recent SARS-CoV-2 variants.

Then to get to that, then it actually uses alpha for multimer to generate a computational predicted structure of the nanobody and the spike protein of SARS-CoV-2 from one of the recent variants.

Another agent is actually recent that okay, so that's a

AI predicted structure, but that could still have some mistakes and biases.

Then it uses like a third step of using Rosetta, which takes us input to that predicted structure, and then

relaxes the structure and then sort of in a more like a physics space way and then estimates the binding energy.

So in the end, there are basically sort of three tools that agent ends ends up using, right?

So the e large language model, AlphaFold, and Rosetta.

And the agents actually decided that okay, you should take do like a weighted score across all three techniques, right?

With a particular sort of weighted scheme, and then iterate this multiple times, right?

So then it does this once, select the not most promising candidates, generate additional mutations and does all three steps again.

And then we'd actually do this four times to generate like a final list of promising nano party candidates.

which are the ones that we then experimentally synthesized and tested.

And that that

People have used the individual tools, right?

But I think using this particular workflow whereas you have this iterative weighted scheme for generating candidates, multiple rounds, uh to the best of our knowledge has not been done in previous publications.

Yeah, so maybe I should have done this earlier, but I want to describe how research happens actually with the virtual lab.

Right.

So as I mentioned before, the virtual lab is really have this set of AI agents.

Showing 101–120 of 371 · page 6 of 19