Tabula: A Single-Cell Foundation Model
film_girl · x · 2026-07-16
This post highlights research on Tabula, a single-cell foundation model that represents cellular data directly as tables rather than artificially constructed gene sequences, and supports privacy-preserving federated training.
The authors claim Tabula not only improves predictive performance but also recovers experimentally validated gene regulatory combinatorial logic across multiple developmental systems. Using an in-house inducible Perturb-seq system, the team found that Oct4 and Sox2 lower the age scores of senescent fibroblasts as predicted. Additionally, candidate factors proposed by the model, such as Cpe, Postn, Olfm2, complement reprogramming factors, expanding the landscape of viable cell reprogramming strategies.
More from coding & agent
- Soft Clamp cuts tool-call overuse in multi-teacher distillation, from 13.7% to 9.0% — antgroup · 2026-07-21
- Agent harness memory loss and compaction are still a major usability problem — adityaag · 2026-07-21
- SpecJudge runs locally on Ollama to pick the right-sized AI model for your project — jokiruiz · 2026-07-21
- A developer maps out six design rules for CLIs that humans and AI agents can both use — yujiezha · 2026-07-21
- A coding-agent skill that forces ADHD-friendly, answer-first output — ayghri · 2026-07-21
- A set of agent skills for CAD, robotics, and hardware design — earthtojake · 2026-07-21