Optimal design of gene knockout experiments for gene regulatory network inference

Area of Science:

• Systems Biology
• Computational Biology
• Genomics

Background:

• Inferring gene regulatory networks (GRNs) from gene expression data is challenging due to underdetermined models.
• Suboptimal experimental design exacerbates the identifiability issues in biological network inference.

Purpose of the Study:

• To develop an algorithm for optimal experimental design in gene knockout (KO) studies for GRN inference.
• To address the underdetermined nature of GRN inference by optimizing KO experiment selection.

Main Methods:

• Developed the REDuction of UnCertain Edges (REDUCE) algorithm.
• Utilized ensemble inference to identify uncertain gene interactions.
• Introduced the concept of edge separatoid for selecting optimal gene knockouts.
• Proposed an iterative procedure combining KO experiments, ensemble updates, and optimal design.

Main Results:

• REDUCE algorithm identifies optimal gene KO experiments for GRN inference.
• Demonstrated efficacy in case studies including Escherichia coli and DREAM 4 100-gene GRNs.
• Achieved higher information return per KO experiment compared to systematic KOs.
• Resulted in more accurate GRN estimations.

Conclusions:

• REDUCE is an enabling tool for underdetermined GRN inference.
• The iterative procedure facilitates efficient and automated GRN inference.
• Advances in technology support the practical application of automated GRN inference.