MICRAT: a novel algorithm for inferring gene regulatory networks using time series gene expression data.

Area of Science:

• Bioinformatics
• Systems Biology
• Computational Biology

Background:

• Gene regulatory network (GRN) reconstruction aims to identify gene-gene regulation relationships.
• High-throughput data from microarrays and RNA-sequencing offer opportunities to infer these relationships.
• Existing GRN inference methods face limitations in accuracy due to data complexity and time delays.

Purpose of the Study:

• To develop a novel algorithm for more accurate GRN inference from time-series gene expression data.
• To address limitations of existing methods by incorporating dependence and time delays in gene regulation.
• To improve the understanding of biological processes through enhanced GRN reconstruction.

Main Methods:

• Developed MICRAT (Maximal Information coefficient with Conditional Relative Average entropy and Time-series mutual information) algorithm.
• Utilized Maximal Information Coefficient (MIC) to construct an undirected graph of gene relationships.
• Directed graph edges using conditional relative average entropies and time-series mutual information to identify regulators and targets, accounting for time delays.

Main Results:

• MICRAT demonstrated improved inference accuracy on synthetic and real gene expression datasets.
• Outperformed existing methods like TDBN on 100-gene networks from the DREAM4 challenge.
• Successfully reconstructed GRNs for the SOS DNA repair pathway and E. Coli datasets.

Conclusions:

• MICRAT is an effective novel algorithm for inferring GRNs from time-series gene expression data.
• The algorithm's ability to account for dependence and time delays enhances its accuracy.
• MICRAT shows superior performance on larger gene networks, offering a valuable tool for systems biology research.