Reconstruction of Gene Regulatory Networks based on Repairing Sparse Low-rank Matrices

Area of Science:

• Systems Biology
• Computational Biology
• Bioinformatics

Background:

• High-throughput proteomic data, especially time-series gene expression, presents challenges in organizing heterogeneous biological information.
• Understanding gene regulatory network (GRN) dynamics under various stimuli is crucial, as biological systems like tumors respond differently to treatments.
• Lack of knowledge in GRN dynamics leads to difficulties in modeling, parameter identification, and structure inference, introducing bias and uncertainty.

Purpose of the Study:

• To develop a novel algorithm for estimating bias error caused by perturbations in gene expression data.
• To correctly identify the common graph structure of gene regulatory networks (GRNs) despite biased inferred structures.
• To retrieve common GRN dynamics across various perturbations, a process termed 'repairing'.

Main Methods:

• A new algorithm is described to estimate bias error resulting from perturbations.
• The method identifies common graph structures by retrieving shared dynamics across perturbed GRNs.
• The approach is inspired by image repairing techniques in computer vision and does not require precise perturbation information.

Main Results:

• The algorithm successfully estimates bias error due to perturbations.
• It accurately identifies the common graph structure among biased inferred GRNs.
• The 'repairing' method automatically corrects the GRN structure across different perturbations.

Conclusions:

• The developed algorithm effectively 'repairs' common GRN structures from heterogeneous, perturbed data.
• This method enhances the accuracy of GRN inference in systems biology.
• The findings have implications for experimental design and interpreting complex biological systems.