Real Alpha

Latent Space · 3 min instead of 33 min · Oct 10, 2026

Messy metagenomic data can improve protein models

SC

Sal Candido

Biohub

  • BA

    Brandon AndersonHost · Moderator

About this segment

Candido says protein language models trained on low-quality metagenomic sequences, many not whole proteins, still design better real proteins. Scaling easy data is a trap, so modelers and data generators must work together in the open.

From “Why AlphaFold Didn't Solve Protein Folding — Pushmeet Kohli, Google DeepMind & Sal Candido, Biohub”, Oct 10, 2026. Watch the full episode

“when we train a protein language model, we train on metagenomic sequences, which are not the highest quality data.”
SCSal Candidoat 3:44 · Latent Space3:44

Takeaways

  1. Noisy metagenomic data can still raise protein model performance
  2. Don't just scale the data that is easy to generate
  3. Open collaboration reveals the data you actually need to find the scaling law
Watch 2:50 from 3:07

In this episode

Full episode

  1. 3:07Messy metagenomic data can improve protein models2:50

Up next · No PriorsFrontier training costs rise 4x per generation, and AI speeds the work4:09

Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

  1. 3:20

    but, like, I'm lazy, so I'm going to work with what's available. And, you know, I think there is actually, like, a good side and a bad side to this. And the good side is that I think when you're doing machine learning or conventional machine learning, you're really looking for the most pristine, high-quality data examples that you can find. But when you actually, if you're like me, you're actually someone who goes and you're kind of like rooting around in the back room looking through people's junk, like what's there.

  2. 3:54

    So a concrete example of this is when we train a protein language model, we train on metagenomic sequences, which are not the highest quality data. In fact, much of that data, I can guarantee you, isn't even a real whole protein. And yet that makes the performance of the model go up for designing real proteins that work, that make for understanding proteins that we know are things. and so that can that's the positive side of it but that can lead you to a negative place where you say let's just scale up the data that we can generate easily and I think that's not the way to do it and I think that's one of the things that's really exciting to me about what we're talking about here today with the VBI it's really going out and saying what is the data that we need to solve the problem.

  3. 4:52

    It's the right resources, but also the right community. I think that one thing that we do or we try and do at Biohub anyway is to work in the open, work with the community and move the whole community forward. And I think that's a really critical piece for this because if you just have people who are building models, we're going to lean towards the data that exists. If you just have people who generate the data, they're going to lean towards the things that can be generated.

  4. 5:23

    But if you can work together as a community every step of the way in an open fashion, then you can actually figure out what is the data that we need, and then you can find that scaling law. Cool. What do you think, Sal? A better lesson for data. Yeah, so I think the... So I was at DeepMind when Rich was with us