Thursday, 9th January, 2020, 12:00

Multi-Omics Factor Analysis: a principled framework for the integration of large-scale multi-omics data

Methods for analysing large-scale multi-omics studies are statistically challenging and currently lacking. I will present MOFA, a statistical framework for the comprehensive and scalable integration of multi-omics data. MOFA builds upon a Bayesian Factor Analysis framework combined with fast GPU-accelerated stochastic variational inference. The model allows for interpreting variation in both bulk and single-cell datasets by pooling information across cells and features to reconstruct a low-dimensional representation of the data. Uniquely, the model supports flexible sparsity constraints that allow joint modelling across multiple groups (conditions or experiments) and views (modalities or assays). To demonstrate the use of MOFA, I will present analysis from data sets of different scales and designs, including personalised medicine applications from patient cohorts and signal extraction from complex single-cell data.

Speaker: Ricard Argelaguet, European Bioinformatics Institute (EBI), Stegle/Marioni Group

Room Aula room 473.10 (4th floor)