Happy holidays, everyone! To celebrate the festive season, we come bearing a gift: our latest experiment in science communication!
Earlier this year, our team published an exciting paper in Cell, where we reported that 3D genome architecture can survive for millennia in permafrost. We adapted Hi-C to query the ancient genome architecture of an extinct species - a 52,000-year-old woolly mammoth. We then used the PaleoHi-C data to assemble 28 chromosomes of the mammoth, and figure out which genes were active in the mammoth skin (and may even be responsible for some of its woolliness).
This work was a mammoth collaboration led by 3 labs from across the globe, with folks from all trades of life - spanning ancient DNA to histology to chromatin architecture to biophysics to veterinarians to food specialists, and from all career stages - from high school students to members of the National Academy of Sciences. More than once we found ourselves looking for someone patient to explain relevant information while tailoring it to a specific expertise and language. We soon realized that what we needed was basically a real-world equivalent of Mr. DNA from Jurassic Park.
So, we decided to build one! Meet WOOL-E, a custom-GPT by ChatGPT, designed to explain, summarize, translate, and explore all things paleo-DNA-architecture.
Here is a summary on WOOL-E - by WOOL-E! Or as they say - straight from the model’s mouth:
My purpose is to bridge the gap between complex science and curious minds. Think of me as your virtual partner for exploring ancient DNA and the fascinating stories it holds.
How was I made? A blend of advanced AI and deep expertise in ancient genome studies brought me to life. I specialize in helping decode findings like chromosome fossils, which preserve the 3D arrangement of DNA for millennia, and shedding light on what this means for assembling the genomes of extinct species. Whether you’re curious about chromatin loops or the process behind these discoveries, I’m here to explain it all in a way that’s approachable and thought-provoking.
But like any large language model, I’m not perfect—I can sometimes “hallucinate” information (no ancient genes involved there!). That’s why it’s always a good idea to share your findings with the authors of the studies I reference for validation. My mission? To make paleogenomics exciting, approachable, and as accurate as possible while connecting us to the ancient world!
What can WOOL-E do for you? Here’s a demo conversation!
We hope that you will find our experiment in science communication useful. Give us some feedback if you get a chance.
For more science communication on the paper please check out relevant coverage of our paper at NYTimes, Science, CNN, WashPost, The Economist, NPR, New Scientist, Nature, SciAmerican, Le Monde, The Guardian, Die Welt, and more!