Gene network interconnectedness and the generalized topological overlap measure.

Area of Science:

• Bioinformatics
• Systems Biology
• Network Science

Background:

• Network methods are crucial for representing gene and protein interactions, suggesting shared biological functions or pathways.
• Existing measures of pairwise interconnectedness can be limited by noisy or incomplete connection data.
• Topological overlap, based on shared neighbors, is a known indicator of functional similarity between proteins.

Purpose of the Study:

• To investigate if higher-order neighborhoods can yield a more robust and sensitive measure of interconnectedness than traditional methods.
• To develop and evaluate a generalized topological overlap measure (GTOM) applicable to m-step neighborhoods (m >= 2).

Main Methods:

• Generalized the topological overlap measure to incorporate m-step neighborhoods (m >= 2).
• Defined the m-th order generalized topological overlap measure (GTOM) by counting shared m-step neighbors and normalizing the value between 0 and 1.
• Applied theoretical arguments and empirical validation using a yeast co-expression network and a fly protein interaction network.

Main Results:

• The proposed GTOM effectively generalizes the concept of topological overlap to higher-order neighborhoods.
• Demonstrated the utility of GTOM in identifying modules and analyzing gene neighborhoods in biological networks.
• The measure provides a normalized value between 0 and 1, facilitating comparisons across different network scales.

Conclusions:

• The m-th order topological overlap measure serves as a valuable filter against spurious or missing network connections.
• GTOM offers a flexible approach to balance sensitivity and specificity in defining pairwise interconnectedness and network modules.
• This generalized measure enhances the robustness and sensitivity of network analysis in biological contexts.