UAE spotlights AI model simulating drug responses

The United Arab Emirates has showcased an artificial intelligence system called AIDO Cell, developed by GenBio AI in collaboration with researchers from Mohamed bin Zayed University of Artificial Intelligence. This virtual-cell "world model" simulates human cellular responses to drugs and other interventions across multiple biological levels, from DNA and RNA to proteins and full-cell behavior.

GenBio AI unveiled a preview of AIDO Cell in August, with its President and University Professor Eric Xing and Professor of Machine Learning Le Song leading the project as founding team members. AIDO Cell is designed to maintain a persistent representation of a cell while interventions are applied, allowing researchers to sequentially simulate gene silencing, overexpression, and compound introduction.

This approach addresses a major challenge in computational biology, as separate specialized models often fail to account for how changes in one biological layer affect others. The system currently integrates multimodal biological data within a shared model, predicting gene expression, RNA splicing, protein structure, and cellular morphology.

GenBio AI emphasizes that the system is still in an early stage, with plans to enhance cell coverage and validate findings through further experimentation. Current demonstrations include simulations of K-562 and Hep-G2 cell lines, examining responses to drug exposure and genetic perturbations. AIDO Cell builds upon previous work on AIDO.

Cell, a single-cell foundation model trained on data from approximately 50 million human cells, which has been evaluated for tasks like cell classification and perturbation modeling. The AI-powered model extends the concept by connecting biological scales within a stateful simulation framework.