Integrating multi-omics data using MOFA2
Integrating multi-omics data using MOFA2
26 November 2026, 13h30-16h30
27 November 2026, 13h30-16h30
General context
This training introduces participants to Multi-Omics Factor Analysis (MOFA2), a statistical framework for the unsupervised integration of multiple omics datasets (Argelaguet et al., 2020). Through this course, you will learn how to run MOFA2 on bulk and single-cell multi-omics data, interpret the factors driving variation across modalities, and apply the results to downstream analyses such as clustering, regression, and classification. The trainer of this session has experience using MOFA2 to integrate datasets from different organisms.
- Prepare and preprocess data for MOFA2 analysis.
- Interpret MOFA2 factors and assess their biological relevance.
- Apply MOFA2 results to downstream tasks such as clustering, regression, and classification.
- Explore options for handling missing values and comparing multiple model fits.
The course consists of an online demo and hands-on session, followed by an interactive Q&A session.
Familiarity with Python.
Those with no Python experience whatsoever are recommended to complete the online Python introduction and Python essentials for data analysis courses.
Those with basic experience in Python who have never worked with Python environments, numpy, pandas, matplotlib, anndata or scanpy should follow the online Python essentials for data analysis course.
This course is part of the learning path Single cell data long reads.
AI methods used
MOFA2 uses an unsupervised machine learning method (factor analysis) to identify hidden factors that explain the main sources of variation across multiple omics datasets. These factors capture both shared and data-specific patterns, enabling integrated analysis and downstream tasks such as clustering, regression, and data imputation.
Trainers
Ricardo Omar Ramirez Flores
Ricardo Ramirez is a Staff Scientist at EMBL-EBI within the Saez-Rodriguez group, where he leads a research line that develops computational methods to study how tissues change in inflammatory and fibrotic diseases using distinct types of molecular data, including bulk, single-cell, and spatial omics. His research focuses on connecting molecular data from patient samples with clinically relevant questions, particularly with long-term prospective biobanks. He has a background in genomics and applied statistics and obtained his PhD in Biosciences from Heidelberg University.