Gap-com: general model selection criterion for sparse undirected gene networks with nontrivial community structure

Area of Science:

• Bioinformatics
• Computational Biology
• Network Science

Background:

• Gene co-expression networks are crucial for understanding biological systems.
• Inferring complex, non-trivial network structures from data is challenging.
• Existing model selection methods may not adequately capture intricate gene interactions.

Purpose of the Study:

• To introduce a novel model selection criterion for sparse complex gene network inference.
• To provide an optimal method for choosing regularization parameters in graphical models.
• To develop a statistic that distinguishes biologically relevant networks from random structures.

Main Methods:

• Formulation of the gap-com statistic, a novel adaptation of the gap statistic.
• Evaluation of expected community counts using data permutations or Erdős-Rényi graph resampling.
• Application of the criterion for selecting regularization parameters in sparse graphical models.

Main Results:

• The gap-com statistic effectively selects gene network models with non-trivial structures.
• Performance evaluation demonstrates superiority over existing methods on simulated and real biological data.
• The criterion successfully identifies clustered genes and hub genes in complex networks.

Conclusions:

• The gap-com statistic is a robust tool for selecting sparse complex gene network models.
• This method enhances the accuracy of gene co-expression relationship estimation.
• It offers a significant advancement in the field of network inference for biological data.