Scripps Research

First, let's find out a bit more about what mitochondria look like and how they operate.

Danielle Grotchen works on understanding mitochondrial biology at the Scripps Research Institute.

If you think of the cell as a fried egg, the yolk would be the nucleus.

And if you looked just around the nucleus, you see these tiny structures that actually form a network within one another.

Nearly seven years after the pandemic began, are we finally piecing together how a viral infection becomes a chronic and disabling disease? Julia Moore Vogel was living with her husband Bobby and their two-year-old daughter in the sun-drenched suburbs of San Diego when the COVID-19 pandemic swept in the United States on commercial airplanes and cruise ships.

Vogel, who is both a PhD in computational biology and medicine, and an MBA from Cornell University, worked for Scripps Research at its facilities in La Jolla.

She cherished her job managing a part of an ambitious study funded by the National Institutes of Health, NIH, known as the All of Us Research Program, which gathers health information from a vast cross-section of Americans collecting biological samples.

health records, and other medical data from nearly 750,000 people that may one day yield treatments for devastating diseases.

So let's start with the brains of the operation, the interpretation layer.

There was this huge experiment run by OpenAI, a startup called Perturb AI and Scripps Research.

Specifically, the Doors Neuroscience Center over at Scripps.

Yeah,

They don't want it to maybe leave their own walls.

I was at an event last week at Scripps Research out here, and the head of OpenAI's life science division was there, and she was talking about how they're training a lot of people training their language model on kind of chemistry, basic science stuff and working with partners to train with their data and then a consulting role.

So it's, I think they understand the data protection necessity that these companies have and they're trying to work around it.

Lily's huge, but yeah, And maybe they have the resources, they'll do it themselves.

Give it read-only access wherever you can and keep a person on anything that moves money.

Now on the flip side, because agents aren't only out there breaking things, On Wednesday, OpenAI, along with a drug discovery company called Perturb AI and Scripps Research, said they ran 1,946 AI agents at the same time on existing data from millions of mouse brain cells, which is kind of wild.

Each agent took one gene and read up on it before it came back with explanations.

And then the team stitched it all together.

If an AI is going to actually understand a disease, it needs to handle that systemic complexity.

Which brings us to the massive scale of GPT-Rosalin, reported on October 23rd, 2026, involving OpenAI, Perturb AI, and Scripps Research.

Yeah, GPT-Rosalin represents a fascinating shift from single molecule design to whole system analysis.

The architecture they deployed ran 1,946 AI agents in parallel.

You are trying to map cascading interactions across millions of cells, where one tiny change triggers a butterfly effect throughout an entire living system.

Which perfectly sets the stage for a massive collaborative experiment between OpenAI, a startup called PerturbAI, and Scripps Research.

Yeah, the scale of what they attempted here is almost hard to

comprehend.

It stops it from launching autoimmune attacks against the body's own tissues or reacting violently to a piece of pollen or a peanut.

And this immune education isn't just a general calm down signal, right? The sources point to a very specific study from Scripps Research that just blew my mind.

Oh, yes.

They were looking at a highly specialized type of immune cell called a mate cell.