decoupleR

decoupleR infers biological activities from omics datasets by integrating prior-knowledge resources and computational methods to quantify regulator activities.

Key Features:

• Unified Framework: Provides a cohesive environment to apply various computational methods for extracting regulator activities from omics data.
• Flexible Methodology: Supports methods that incorporate modes of regulation and interaction weights to tailor analyses to specific network properties.
• Integration with OmniPath: Leverages OmniPath, a meta-resource composed of over 100 databases of prior knowledge, to inform activity inference.
• Performance Evaluation: Evaluation on transcriptomic and phospho-proteomic perturbation experiments showed that simple linear models and consensus scores across top-performing methods are effective at predicting perturbed regulators.
• Prior-knowledge dimensionality reduction: Uses prior-knowledge integration to reduce dimensionality, enhancing statistical power and interpretability of inferred activities.

Scientific Applications:

• Transcriptomics and Phospho-Proteomics: Identifies changes in gene expression and protein phosphorylation indicative of underlying regulatory mechanisms in perturbation experiments.
• Predictive Modeling of Perturbed Regulators: Predicts perturbed regulators to support studies of genetic or environmental interventions on biological systems.

Methodology:

Applies a range of computational methods—including simple linear models and consensus scoring—using prior-knowledge integration to reduce dimensionality and infer regulator activities from omics data.