Detection of multiple perturbations in multi-omics biological networks

Area of Science:

• Computational biology
• Systems biology
• Genomics

Background:

• Understanding cellular mechanism-of-action is crucial for disease identification and treatment development.
• Pinpointing the origin of cellular disturbances is challenging due to complex downstream effects.
• High-throughput biological data offers insights but poses analytical challenges for multi-source data integration.

Purpose of the Study:

• To develop a novel computational method for mechanism-of-action inference using multi-attribute biological data.
• To extend network filtering techniques to integrate diverse data types for perturbation analysis.
• To accurately identify the primary site of cellular perturbations and detect multiple disturbances.

Main Methods:

• Estimation of a joint Gaussian graphical model across multiple data types via penalized regression.
• Application of network filtering to identify significant network effects.
• Utilizing likelihood ratio tests and a conditional testing procedure for perturbation identification.

Main Results:

• The proposed method effectively infers cellular mechanism-of-action from multi-attribute data.
• Demonstrated capability to identify the most likely site of original perturbation.
• Successfully detected multiple perturbations through a conditional testing procedure.

Conclusions:

• The developed methodology provides a robust framework for mechanism-of-action inference.
• This approach enhances the analysis of complex biological systems using integrated multi-omics data.
• The method shows promise for applications in disease research and therapeutic development using datasets like TCGA.