A matrix rank based concordance index for evaluating and detecting conditional specific co-expressed gene modules

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene co-expression network analysis (GCNA) is crucial in bioinformatics for predicting gene function, inferring protein interactions, identifying disease markers, and discovering copy number variations.
• A significant gap exists in understanding the mathematical conditions governing co-expressed gene modules.

Purpose of the Study:

• To introduce a novel linear algebraic method, the Centralized Concordance Index (CCI), for evaluating the concordance of gene modules identified through GCNA.
• To establish CCI as a tool for assessing GCNA algorithm performance and detecting condition-specific co-expression modules.

Main Methods:

• Developed a linear algebraic approach to compute the Centralized Concordance Index (CCI).
• Applied CCI to evaluate gene modules in the context of gene co-expression network analysis.
• Utilized CCI for the specific task of detecting lung tumor-associated gene modules.

Main Results:

• Simulations demonstrated CCI's robustness as an indicator for assessing the concordance of co-expressed gene groups.
• Application to lung cancer datasets identified potential tumor-specific genetic alterations, including copy number variations (CNVs) and potential gene fusions.
• CCI proved more resilient to outliers and interfering modules compared to density-based methods using Pearson correlation coefficients.

Conclusions:

• The Centralized Concordance Index (CCI) effectively evaluates GCNA algorithm performance and identifies condition-specific co-expression modules.
• CCI offers a robust alternative to existing methods, showing improved performance in the presence of noise and outliers.
• Further investigation is warranted to elucidate the molecular mechanisms underlying condition-specific co-expression relationships identified by CCI.