MPAC: a computational framework for inferring pathway activities from multi-omic data.

Area of Science:

• Computational biology
• Bioinformatics
• Systems biology

Background:

• Understanding cellular states requires integrating multiple omic data types (genomic, transcriptomic, proteomic).
• Multi-omic data analysis is crucial for disease research, potentially revealing novel patient subgroups and informing treatment strategies.
• Existing methods may struggle to reconcile complex and conflicting measurements from diverse omic assays.

Purpose of the Study:

• To present Multi-omic Pathway Analysis of Cells (MPAC), a computational framework for interpreting multi-omic data.
• To leverage biological pathway knowledge to infer consensus activity levels of cellular entities.
• To group biological samples based on pathway activities and identify proteins with clinical relevance.

Main Methods:

• MPAC utilizes factor graphs to encode network relationships from biological pathways.
• It infers consensus activity levels for proteins and pathway entities from multi-omic data.
• Permutation testing is employed to eliminate spurious activity predictions, and samples are grouped by pathway activities.

Main Results:

• MPAC successfully predicted a patient subgroup related to immune responses in head and neck squamous cell carcinoma, which was not identified by single-omic analyses.
• Key proteins identified within this subgroup showed associations with clinical outcomes and immune cell compositions.
• The framework demonstrates the power of integrating multi-omic data with prior biological knowledge.

Conclusions:

• MPAC offers a robust computational approach for multi-omic data interpretation, enhancing disease subgroup discovery.
• The framework prioritizes proteins with potential clinical relevance by linking pathway activities to patient outcomes.
• The MPAC R package facilitates advanced multi-omic analyses for researchers.