EdHarris

I’m not sure why people are getting so worked up about this. Ordering flowers for someone abroad can be surprisingly difficult. After staying with someone in the US, I tried to send them flowers as a thank-you once I was back in the UK. For some reason, it was an absolute pain. If this makes that easier, great.

I also doubt my mum (or anyone else I’m buying flowers for) would care whether I ordered through a website or asked a bot to help. The thoughtful part is remembering them and deciding to send flowers. I’m not sure anyone yearns for me to personally hunt down a website and type in my credit card details.

And people already love flower subscriptions. Aren’t those essentially automated flowers?

Folding@home uses Rosetta, a physics-based approach that is outperformed by deep learning methods such as AlphaFold2/3.

Yes, all data Google used was public. We have enough compute from YC (thanks YC!) to do this. The main thing is the technical infrastructure - processing the data, efficient loading at training time, proper benchmarking, etc. We are building these now.

Thanks!

The predictions can be verified by comparing the predicted structure to the experimentally solved structure, either crystal or cryoEM. The model is still training and improving, we will release the benchmarking results after it's complete.

Amazing! What kind of things did you work on?

Yes this is a good point. We are actively speaking with our counsel to check this. Thanks for flagging, though.

OpenFold, which was AlphaFold2's open-source implementation was published in Nature Methods. We will prepare a similar publication once the model is more mature and when we have a nice set of experiments showing the model's interesting properties.

We think enzymes are super cool! You can build molecular assembly lines at the atomic scale with them. Many pharmaceuticals are already manufactured with enzymes such as the diabetes drug Januvia. Engineering them is a big bottleneck though - takes years and millions of dollars. We want to speed this up with AI-powered design. Next step is ligand-protein prediction capability of AlphaFold3, which is also super useful for modelling enzyme-substrate interactions.

Our long term goal is to design enzymes for chemical manufacturing. We decided to build AlphaFold3 because we had seen how useful AlphaFold2 had been for the protein design field. No one else was building it fast enough for us, so we decided we should do it ourselves. We are committed to training and open-sourcing the full version with ligand and nucleic acid prediction capabilities as well since it is so useful for the biotech industry.