CZI Science
@cziscience
Followers
27K
Following
5K
Media
2K
Statuses
6K
Science news and updates from the @ChanZuckerberg
SF Bay Area
Joined June 2017
šØ Just dropped! GREmLN is a graph-based transformer model that uses the molecular logic of cells to enable insights into disease states. This nimble model accounts for causality + locality using gene regulatory networksš Read the preprint:
0
8
33
Massive cross-disciplinary teams (biologists, engineers, data scientists) are driving new modes of imaging + analysis. Itās not just tech, itās collaboration for discovery
0
2
14
RT @Radiolab: Can AI find lifesaving cures hidden in plain sight?. Listen to āThe Medical Matchmaking Machineā here: .
0
1
0
RT @czbiohub: š£ Accurate cell tracking remains one of the hardest problems in #bioimaging. Meet Ultrack, a fast, reliable platform that perā¦.
0
6
0
Dig into the basics of virtual cells with this @freethink primerāAI-driven models, applications in research, & the challenges weāre tackling. Read more:
freethink.com
Biologists are skipping the petri dish and using AI-powered virtual cells to experiment in silico.
0
2
9
šWeāre proud to support the2025 Global Science Scholars!. Six postdocs from the U.S. are heading to Japan; four from Japan to the U.S. Together, theyāre advancing cutting-edge research in biology. #CZBiohubSF #CZBiohubNY #CZBiohubCHI
czbiohub.org
Apply for the new international postdoctoral fellowship program from the CZ Biohub Network and Stellar Science Foundation.
1
2
7
In the future, scientists could simulate experiments or brainstorm new therapies just by chatting with AI. Itās all part of our effort to bring the virtual cell to life, empowering researchers around the world. š Model š Preprint
biorxiv.org
Reasoning Models are typically trained against verification mechanisms in formally specified systems such as code or symbolic math. However, in open domains like biology, we do not generally have...
0
0
1
These AI assistants for biology are:. ā¶ļø More understandable to humans.ā¶ļø Flexible across different kinds of biological questions.ā¶ļø Grounded in real science, not just language mimicry
chanzuckerberg.com
Scientists can ask complex biological questions in plain language and get predictions about gene interactions.
1
0
1
Key results:.ā¶ļøSimulation accuracy is competitive with experimental data.ā¶ļøVirtual Cell Models teach rBio off-task biology exhibiting transfer to perturbation.ā¶ļøBiological databases can train the model + mixing datasets helps it generalize.ā¶ļøWe can prompt rBio in natural language.
1
0
1
šØ Just released: rBio, the first LLM trained with virtual cell models to perform biological reasoning. This is a big step toward building AI-powered virtual cells that help predict + understand how cells behave in health & disease š§µā¬ļø
biorxiv.org
Reasoning Models are typically trained against verification mechanisms in formally specified systems such as code or symbolic math. However, in open domains like biology, we do not generally have...
1
4
17
New to virtual cells? @freethinkās primer explains these AI-powered, in silico modelsāwhy theyāre game-changers for biology & the hurdles still to solve. Worth a read
freethink.com
Biologists are skipping the petri dish and using AI-powered virtual cells to experiment in silico.
0
7
16
Collaboration, not competition, will map the future of human biology. The Billion Cell Project brings together 10 institutions, over 100 experts, 10 research centers + a shared vision: unlocking biological insight at #SingleCell resolution.
genengnews.com
As single-cell RNA-Seq meets AI, and DNA sequencing prices drop, projects are becoming much (much) larger.
1
36
149
The new CZI Biohub Imaging Institute/SF Biohub collab is pushing the limits of in vivo imaging. Think multiscale, dynamic maps of biology in action. A big leap for cell & developmental biology!
chanzuckerberg.com
Collaboration is changing whatās possible in imaging science
0
2
10
š The @Falling_Walls Science Breakthroughs of the Year is one of the worldās most inspiring recognitions in science. Weāre so proud of David Maresca + Shana Kelley, 2025 finalists for their visionary work.
0
0
4
Curious about #ComputationalBiologyās next leap?.@freethinkās new primer unpacks virtual cells: AI-driven models simulating cell biology in silico. Why build them? For scalable hypothesis testing, drug design + systems-level insights.
freethink.com
Biologists are skipping the petri dish and using AI-powered virtual cells to experiment in silico.
0
1
4
Sharing negative results = saving time for other researchers. @dshugo at CBMRT is creating an #OpenScience AI-powered tool, the Null Research Compass, to make it easier for scientists to publish + find these results.
0
0
3
RT @freethinkmedia: What if AI virtual cells could help us decode disease, test medication, and rewrite biology? . Read the latest editionā¦.
0
29
0
RT @czbiohub: "Although there are some precedents to this, the new system is unlike anything before it," writesĀ @labriataphd inĀ @TDataScienā¦.
towardsdatascience.com
Welcome to the 21st century by the hand of large language models and reasoning AI agents
0
3
0
The Billion Cell Project isnāt just about data volumeāitās about creating the reference scaffolds for models of human biology. We see this as foundational to building virtual models for discovery, disease modeling + therapy development.
genengnews.com
As single-cell RNA-Seq meets AI, and DNA sequencing prices drop, projects are becoming much (much) larger.
0
7
23
RT @czbiohub: Double paper alert in @Nature! A team of 150+ scientists, led by @StephenQuake, Mark Krasnow/@Stanford & #CZBiohubSF Genomicsā¦.
0
4
0
Our seed oil-free chips are made with just three ingredients:. Potatoes,.Beef Tallow,.And Sea Salt. Order yours today to taste the tradition. š¤ .
0
64
359
#CZBiohubNY + @ChanZuckerberg invited a cross-disciplinary group of minds to rethink core biological questions & dive into:.š§ In vivo perturbations.š§¬ Sequence-to-function models.š¬ Organelle engineering.š Multi-scale dynamics. more!
2
3
51