Inferring gene networks: dream or nightmare?

Area of Science:

• Computational Biology
• Systems Biology
• Bioinformatics

Background:

• The Dialogue for Reverse Engineering Assessments and Methods (DREAM2) competition provided a platform for evaluating gene regulatory network inference algorithms.
• Previous methods for analyzing biological data, including perturbation and time-series data, required further refinement for complex challenges.
• Genome-scale networks, such as that of Escherichia coli, present significant analytical challenges due to data complexity and noise.

Purpose of the Study:

• To present and evaluate algorithms with winning performance in the DREAM2 Reverse Engineering Competition 2007.
• To investigate methods for improving algorithm performance on in-silico challenges using perturbation and time-series data.
• To assess the effectiveness of various association measures for analyzing genome-scale biological networks.

Main Methods:

• Development and application of novel algorithms for gene regulatory network inference.
• Explicitly addressing perturbation and time-series data for in-silico challenges (Challenge 4).
• Evaluation of multiple association measures on a genome-scale Escherichia coli network dataset (Challenge 5).

Main Results:

• Algorithms achieved winning performance in the DREAM2 competition.
• Modifications to original methods led to substantial performance improvements for in-silico challenges.
• Analysis of the Escherichia coli network was challenging; best results were obtained by analyzing sub-datasets.

Conclusions:

• Algorithm performance in gene network inference can be significantly improved through method refinement and tailored data handling.
• Genome-scale network analysis, particularly with complex datasets, remains a challenging area in systems biology.
• Sub-dataset analysis can be a viable strategy when dealing with large, intricate biological networks.