Speaker
Description
The advent of temporal single-cell RNA sequencing (scRNA-seq) data has enabled in-depth investigation of dynamic processes in heterogeneous multicellular systems. Despite remarkable advancements in computational methods for modeling cellular dynamics, integrating cell-cell interactions (CCIs) into these models remains a major challenge. This is particularly true when dealing with high-dimensional gene expression profiles from large populations of interacting cells, where the intricate interplay between cells can be obscured by data complexity.
In this talk, I will present scIMF, a deep generative Interacting Mean Field model that learns collective multicellular dynamics directly from temporal scRNA-seq data. Built on the McKean-Vlasov stochastic differential equation (MV-SDE) framework , scIMF models each cell's dynamics as a function of both its own gene expression state and the empirical distribution of the entire cell population. A cell-wise Transformer attention mechanism parameterizes the interaction term, enabling efficient inference of nonlocal and asymmetric CCIs.
Benchmarked against state-of-the-art methods across zebrafish embryogenesis, mouse fibroblast reprogramming, and pancreatic β-cell differentiation datasets , scIMF achieves superior gene expression reconstruction at unobserved time points and more accurate cellular velocity inference. Furthermore, scIMF uncovers biologically interpretable, non-reciprocal interaction patterns of cells, providing a principled framework for studying complex, particularly non-equilibrium biological systems.
