A scalable algorithm for structure identification of complex gene regulatory network from temporal expression data.

Area of Science:

• Computational Biology
• Systems Biology
• Genomics

Background:

• Gene regulatory interactions are crucial for biological functions.
• Existing computational methods struggle with genome-wide regulatory landscapes in complex eukaryotes and large networks.
• The curse of dimensionality limits current approaches for networks exceeding 100 nodes.

Purpose of the Study:

• To develop a novel, scalable algorithm for identifying genome-wide gene regulatory network (GRN) structures.
• To overcome the limitations of existing methods in handling large and complex biological networks.
• To provide a tool applicable to large-scale GRN identification in eukaryotes.

Main Methods:

• Developed a scalable algorithm incorporating prior biological knowledge and network topological properties (sparsity, hub structure).
• Utilized a regularized formulation to handle complex network characteristics.
• Verified algorithm performance through extensive simulations using DREAM challenge benchmark data.

Main Results:

• The algorithm demonstrates superior performance for networks up to 10,000 nodes, overcoming scalability issues.
• Validated the algorithm using time-course gene expression data from human respiratory epithelial cells infected with influenza A virus (IAV).
• Integrated CHIP-seq data from ENCODE to analyze transcription factor (TF) and target gene interactions, revealing key TFs in IAV infection.

Conclusions:

• The proposed algorithm is the first scalable method for identifying large, complex GRN structures.
• Identified GRN structures can reveal biological links and guide functional investigations.
• The algorithm is implemented in MATLAB and the source code is publicly available.