MultiCellFM: Learning Meaningful Representations of Single Cell Data

AI Technology: Natural Language Processing; Deep Learning. 

MultiCellFM aims to develop an advanced Cell Foundational Model (CFM) that leverages cutting-edge artificial intelligence to extract meaningful insights from vast single-cell datasets. 

The project integrates Perceiver IO, an architecture optimized for multimodal data processing, with scPerturb, a data harmonization framework tailored for large-scale single-cell perturbation studies. 

By unifying transcriptomics, proteomics, and epigenomics in a single AI-driven framework, MultiCellFM eliminates the need for modality-specific feature engineering, improving the generalizability and scalability of cell-based models. 

This approach will enable predictive modeling of cellular responses to small molecule perturbations, a critical challenge in drug discovery and precision medicine.