MOFA-FLEX
MOFA-FLEX – Flexible Bayesian Factor Analysis for Omics Data
MOFA-FLEX is a versatile factor analysis framework designed to streamline the construction and training of complex matrix factorization models for omics data. It is built on a probabilistic programming-based Bayesian factor analysis framework that integrates concepts from multiple existing methods while remaining modular and extensible. MOFA-FLEX generalizes widely used matrix factorization tools by incorporating flexible prior options, including structured sparsity priors for multi-omics data and covariate-informed priors for spatio-temporal data, as well as non-negativity constraints and diverse data likelihoods. This allows users to combine model components according to their specific analysis needs.
Key benefits
Combines concepts from many published matrix factorization methods in one package
Allows flexible combination of model components, priors, constraints, and likelihoods
Naturally handles missing data within a Bayesian modelling framework
Supports interpretable latent factors that summarize sample structure and identify driving molecular features
Enables fast inference using GPUs
Applications
Exploratory analysis of complex omics datasets
Multi-omics integration of bulk and single-cell data
Spatio-temporal modelling using continuous covariates across space or time
Pathway-informed analysis using known gene sets and biological prior knowledge
Semi-supervised and guided factor analysis for technical covariates and CRISPR perturbation experiments
Intended use
MOFA-FLEX is intended for bioinformaticians, computational biologists, systems biologists, single-cell researchers, and multi-omics researchers who need flexible matrix factorization models for complex omics datasets. It is particularly suited for users who want to combine features from different factor analysis methods, analyse bulk or single-cell data, incorporate prior biological knowledge, or model structured datasets such as spatio-temporal data and CRISPR perturbation experiments.
Contact:
Website