Reproduce “AI Scientists for Building Virtual-Cell Models” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Xiangru Tang, Jiapeng Chen, Daniel Shao, Zhuoyun Yu, Yan Cui, Weixu Wang, Fang Wu, Yuchen Zhuang and 7 more

CellForge turns the task of predicting cellular responses to perturbations into a trainable, testable model. Its agents use context to design, implement, and evaluate models against held-out perturbations, with human checks on plausibility and code compliance. On Adamson single-gene CRISPRi, CPA-X achieved the highest compared global profile correlation, 0.9883, but had higher differential-expression error than the best reference methods.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.