Multi-Omics Factor Analysis Plus: A probabilistic framework for comprehensive and scalable integration of multi-modal data

Technological advances have led to a growing number of studies that profile multiple molecular layers simultaneously on large cohorts of samples (e.g., patient tissues) and, more recently, at single cell resolution. While this promises a more comprehensive understanding, computational strategies for an integrative analysis are essential to obtain valuable insights into the underlying biological processes. Furthermore, such methods need to handle complex experimental designs that include multiple data modalities and multiple groups of samples as well as scale to the growing dimensions of the data. We present Multi-Omics Factor Analysis (MOFA) and its extension MOFA+ that provide a statistical framework for the comprehensive and scalable integration of such multi-modal data. MOFA+ reconstructs a low-dimensional representation of the data using a fast stochastic variational inference scheme and employs flexible sparsity constraints that enable joint modelling of variation across multiple sample groups and data modalities. It enables decomposing variation into factors present in all, some, or single data modalities or sample groups and promotes interpretable factors that can directly be linked to molecular drivers. Once learnt, the factors enable a variety of downstream analyses, including the inference of differentiation trajectories, denoising, feature selection or the identification of sample subgroups.