Gene Regulatory Networks Reconstruction Using the Flooding-Pruning Hill-Climbing Algorithm

Area of Science:

• Computational Biology
• Genomics
• Systems Biology

Background:

• The increasing volume of genomic data offers potential for improved gene regulatory network (GRN) reconstruction.
• Bayesian networks are promising for GRN reconstruction due to their probabilistic nature.
• Existing Bayesian network methods face challenges with excessive computation time and large sample size requirements.

Purpose of the Study:

• To introduce the Flooding-Pruning Hill-Climbing (FPHC) algorithm, a novel hybrid Bayesian network approach for GRN reconstruction.
• To address the limitations of existing methods, particularly in scenarios with limited biological samples.

Main Methods:

• Development of the FPHC algorithm, integrating Bayesian networks with a novel DPI Level concept based on data processing inequality (DPI).
• Utilizing a search-and-score approach within a restricted search space for network structure learning.
• Theoretical analysis and validation of FPHC effectiveness.

Main Results:

• FPHC demonstrates superior performance compared to standard hill climbing and Max-Min Hill-Climbing (MMHC) in terms of network structure and running time.
• The algorithm effectively identifies gene neighbors even with limited biological samples.
• Extensive experiments on known Bayesian networks and DREAM challenge datasets validate FPHC's efficacy.

Conclusions:

• The FPHC algorithm offers a more efficient and accurate solution for gene regulatory network reconstruction.
• FPHC is particularly well-suited for applications involving limited biological data, overcoming a key limitation of traditional methods.