PathIntegrate

PathIntegrate integrates multi-omics data at the pathway level to generate interpretable pathway-centric predictive models for predicting biological outcomes and identifying key pathways and molecules relevant to disease mechanisms.

Key Features:

• Pathway-Level Transformation: Performs single-sample pathway analysis to convert molecular-level multi-omics data into pathway-level representations.
• Predictive Modeling: Implements single-view and multi-view predictive models and ranks multi-omics pathways by their contribution to outcome prediction.
• Layer-Specific Contributions: Quantifies the contribution of each omics layer (e.g., genomics, proteomics) to the overall model.
• Molecular Importance Identification: Identifies and ranks the importance of individual molecules within pathways.
• Signal Detection in Low Signal-to-Noise Scenarios: Detects biological signals in datasets characterized by low signal-to-noise ratios.
• Low Effect Size Identification: Identifies significant pathways exhibiting low effect sizes.

Scientific Applications:

• Chronic Obstructive Pulmonary Disease (COPD): Integration of high-dimensional multi-omics datasets to derive insights into disease mechanisms and potential therapeutic interventions.
• COVID-19: Integration of multi-omics data to identify pathways and molecules implicated in COVID-19 biology and outcomes.

Methodology:

Computational methods include single-sample pathway analysis to transform molecular data to pathway-level features; construction of single-view and multi-view predictive models; ranking pathways by contribution to predictions; quantifying omics-layer-specific contributions; and identifying and ranking individual molecules within pathways.