Inside a Nobel Prize Winner’s Lab Where Microscopes Meet AI
7 minAudio
Most microscopes have a limit to what they can show us. Eric Betzig helped to break that limit, and won a Nobel Prize in the process. In this episode, Méline goes behind the scenes at Eric’s lab at UC Berkeley. Eric is a physicist and a winner of the 2014 Nobel Prize in Chemistry. He shows her a microscope built to look inside transparent zebrafish, where blood cells can be seen flowing through capillaries live. He explains why he believes most biology is done on the wrong cells. He also explains how his Cell Observatory initiative is using AI to understand life as a whole system rather than a collection of parts. Along the way, Méline sees the first microscope Eric designed for Eikon Therapeutics, and a view inside the eye of a living fish. He explains: ◼ Why studying cultured cells on glass can tell us little about what happens inside a living body ◼ Why zebrafish are used to understand human biology ◼ How the lab is pushing microscopes to capture data faster than ever before ◼ Why AI and computing power are the next bottleneck in biology ◼ Why he believes observation, not hallucination, is the way forward 0:00 Intro 0:31 Breaking the Limit of the Microscope 0:45 A Microscope Built to Look Inside Transparent Fish 1:31 Watching Blood Flow Through a Living Fish 1:49 Why Most Biology Studies the Wrong Cells 2:32 Why Zebrafish 2:48 Pushing the Speed Limits of Data 3:20 The Cell Observatory and AI 4:03 What Eric Needs From the AI World 5:04 The Microscope He Designed for Eikon 6:00 Inside the Eye of a Zebrafish 6:22 Observation, Not Hallucination ◼ Eric Betzig, UC Berkeley: https://physics.berkeley.edu/people/faculty/eric-betzig ◼ Eikon Therapeutics: https://www.eikontx.com
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
SPEAKER_02Eric Bestig is a Nobel Prize winning physicist. Today we're at his lab at UC Berkeley to get a look behind the scenes, explore the technology he and his team are developing, and find out what happens when you push the boundaries of what we can see.
SPEAKER_00Everything is so complicated in science and so difficult. When you get your experiment working and it's getting data, there's just no rush better than that.
SPEAKER_02Most microscopes have a limit to what they can show us. Eric Betsig helped break that limit, winning a Nobel Prize in the process. Today I'm at his lab at UC Berkeley to see the technology up close.
SPEAKER_00So this is a microscope to look at transparent fish, which is our model organism that we use to understand life. This is the imaging camera. The fish itself is right inside these little wells inside of there. Okay. So there's two lenses here that get opposite images of a fish that's lined up this way. So everything that you see here is all about seeing something inside of there that might be about this big. So it's basically some a light sheet microscope, an adaptive optical microscope, and a fluorescence microscope.
SPEAKER_00This is a very specialized microscope to do the kind of imaging we need to understand what we need. Let's go to this other room now, which is where things are ah yes, it's actually happening as we speak. Those are actually uh blood cells. And this is a little uh capillary, and the and the these are two capillaries, and the blood is flowing through the fish live right there. Okay. So a lot of what you were seeing right there was some blood cells there. Almost all biologists study what are known as cultured cells.
SPEAKER_00They're cells that have been taken out of an organism, usually human. Their DNA has been modified in a way that they can live outside of the body, and they can live basically attached to a glass cover slip. But they've done so much to change them so that they can live that way, that even if your microscope is perfect, you can see every molecule, you're looking at artificiality. The vast majority of biology that's done by images, by looking at immortalized cells and cover slips, there is no reason to believe that what you see there will be translated to what you see inside of a whole living organism.
SPEAKER_00And nor and so because this or because this fish is a vertebrate, like you and me, it's fairly high up on the evolutionary tree. So there's it has a lot of, in fact, 70% of its genes are basically the same genes that we have. The purpose of this scope is really trying to push the speed limits of how much data we can get, because the in order to be able to understand biology as we want to do it, our our estimates of how much data we need to train AI models is a mess.
SPEAKER_00And so in order to get that in any rational amount of time, like a few years, we have to have microscopes that are far more performative than ever before. And that's what this microscope is about.
SPEAKER_02I meet some of Eric's team around the lab, but this is about more than just microscopes. Through Eric's initiative, the cell observatory, they're using AI to rethink how we study life, looking at the bigger picture rather than just the individual parts.
SPEAKER_00The fundamental unit of biology is the cell. And when you're looking at something like you or a zebrafish, there's lots and lots of cells there. And when you do that imaging, like I say in 5D, it all just looks like a smush, you don't even know where one cell begins and another cell ends. What we want to do is train an AI model that can do that at the scale, at speed at which we take this data. They can spread all those cells apart so we can operate them in a virgin.
SPEAKER_02If you are looking for the best outcome, the type of person or people you want to work with, who that will be and how you're going to you know explain clearly how you can cover it with a company.
SPEAKER_00From our perspective, we're pretty vertically integrated on everything. We have a good young small AI team. We would definitely benefit from dealing with you know people who are at the top of the field in AI, but in particular, we need compute. The compute is very hard to come by now because there's such demand for it. It's hard to build an AI model if you don't have the GPUs to build it, right? And so that's that's number one. But we would certainly benefit from discussions about architectures for AI and uh other ways of optimizing what we want to do for what we want to do.
SPEAKER_00So we would love to get more AI people involved. So this microscope is um basically the very first microscope of the type that I designed for Icon. And so Icon uses this as their primary assay for drug discovery. So molecules are moving randomly, but the thing that makes things become ordered is the fact that only certain molecules will stick to one another, right?
SPEAKER_00And so being able to modulate that stickiness can either cause a disease or help cure a disease. So this is basically the basis of what Icon is looking at is trying to understand how the motion and sticking of these molecules is affected when you put in one of a library of maybe a million different drug compounds, and then try to find out which ones do the type of recovery of the kinetics that you want to have for a healthy outcome.
SPEAKER_00Right here, you see this? This is the lens in the eye of that fish. And there's all these cells, the really thin cells that wrap around to create this high refractive index thing that acts like glass to act as the eye. And then these are different um receptors in the retina around the eye of the fish, right? So much biology is hallucination from assuming from a very sparse data set what's actually going on. We believe observation rather than hallucination is the way forward, right?
SPEAKER_02What I've learned is that seeing more is only half the story. By combining powerful microscopes with AI, Eric and his team are working towards a much bigger picture of how life works and how all its moving parts connect.
SPEAKER_01Thank you for watching this new episode of Humanware Project. So that we can bring out more amazing content like this one to you. Please like and subscribe. Your support means the world to us.
At 28, Richard Branson bought an island he couldn’t afford. Decades later, it is still his home, and the place where he lives by the principles that built the Virgin Group. In this episode, Méline returns to Necker...
As a young teenager, Randy Schekman cycled to his local police station and told the duty officer he was running away from home. His mother kept borrowing the money he was saving for a microscope. Half a century later,...
Eric Betzig once walked away from science altogether. Years later, he and his best friend put $25,000 each of their own money into building a microscope on a living room floor, and it won him the Nobel Prize. In this...
What if some of the best health and longevity secrets are not found in laboratories or supplements, but growing on trees right in front of us? In this episode, Méline travels to Hainan, China’s tropical island, with her...
At Golden Door, one of America’s best-known health retreats, guests are served a light tomato broth in the middle of the day, after a morning of hard work and before lunch at one o’clock. It looks simple, but there is a...