CZI @ STAT Breakthrough Summit: Virtual cells might be closer than we think — thanks to AI

Katie Palmer, Gita Mahmoudabadi, and Patricia Brennan speak on stage at the 2025 STAT Breakthrough Summit West, seated in front of a large blue screen displaying the event logo.
STAT reporter Katie Palmer speaks with CZI’s Gita Mahmoudabadi and Patricia Brennan about how CZI is building and training AI models to accelerate biomedical research. Photo by Sarah Gonzalez for STAT

At STAT 2025 Breakthrough Summit West, Patricia Brennan, vice president of science technology and general manager for science; and Gita Mahmoudabadi, senior research scientist, AI; joined STAT reporter Katie Palmer for a fireside chat on how CZI is building and training AI models to accelerate biomedical research. Their conversation focused on the release of TranscriptFormer, a new AI model that is trained on 112 million cells from 12 different species, covering 1.5 billion years of evolution and representing the most evolutionarily diverse corpus of training data. Scientists will be able to use the model to predict what different types of cells are, whether a cell is diseased and how genes interact. TranscriptFormer is an important step toward CZI’s goal of building an AI-based virtual cell model to predict and understand cell behavior.

Learn more about TranscriptFormer and our scientific grand challenge to build AI-powered virtual cells to help scientists unlock the inner workings of human biology and move closer to our goal to cure, prevent and manage all diseases in this century.


###

About the Chan Zuckerberg Initiative
The Chan Zuckerberg Initiative was founded in 2015 to help solve some of society’s toughest challenges — from eradicating disease and improving education, to addressing the needs of our local communities. Our mission is to build a better future for everyone. For more information, please visit chanzuckerberg.com.

News

  • The Scientist: Three Amino Acids Improve Lipid Nanoparticle Therapy Delivery to Cells

  • Simple ‘Cocktail’ of Amino Acids Dramatically Boosts Power of Anti-Inflammatory mRNA Therapies and CRISPR Gene Editing

    Adding three common amino acids to lipid nanoparticle injections increased mRNA delivery up to 20-fold, pushed gene editing efficiency to nearly 90%, and suppressed inflammation in a model of acute liver disease.

  • New tool reveals how T cell responses evolve across organs

    By tracking recently activated T cells over time and across tissues, researchers uncover immune dynamics that may inform future therapies for infection, cancer, and autoimmunity