COGENT

COGENT provides methods to construct and evaluate gene co-expression networks from gene expression data within an R package, guiding selection of co-expression measures and score thresholds using only intrinsic dataset properties.

Key Features:

• R implementation: Distributed as an R package for analysis of gene co-expression networks.
• Flexibility in co-expression measures: Supports multiple statistical measures including Pearson and Kendall correlation coefficients.
• Threshold determination: Assists in selecting score cut-offs for network construction to define significant connections.
• Intrinsic, data-driven evaluation: Compares network construction methods based on the inherent characteristics of the dataset without requiring external validation data.
• Applicability to similarity-profiling data: Can be applied to network construction tasks that are based on similarity profiling beyond standard gene expression datasets.

Scientific Applications:

• Gene expression analysis: Construction and evaluation of gene co-expression networks to study coordinated gene expression patterns.
• Microbiome studies: Application of co-expression/similarity-based network construction to microbial abundance or feature profiles.
• Synthetic lethality analyses: Use in constructing similarity-based networks relevant to synthetic lethality investigations.

Methodology:

Evaluates different co-expression measures (e.g., Pearson, Kendall), assesses score cut-offs/thresholds, and compares network construction methods using intrinsic dataset properties without external benchmarks.