wTO

wTO computes weighted topological overlap scores to quantify pairwise similarity in gene co-expression networks by incorporating shared network-neighborhood information to reduce spurious correlations, particularly in low-sample gene-expression datasets.

Key Features:

• Weighted Topological Overlap measure: Incorporates information from the shared network-neighborhood of gene pairs into a single wTO score.
• Spurious correlation mitigation: Targets detection and reduction of spurious correlations or links when only limited measurements per gene are available.
• Robustness versus simple correlation: Produces more robust scores than simple correlation calculations in simulated low-sample ensembles.
• Performance on small sample sizes: Was a better predictor of full-data correlation scores for data sets with 10 and 20 samples per gene when compared to a 338-measurement reference.
• Soft-threshold modifier effect: Applying a soft-threshold modifier to link weights before computing wTO decreases robustness but increases predictive power relative to the reference correlation network as data set size increases.
• Network topology impact: Using wTO as a score substantially alters several topographical aspects of the resulting networks compared to traditional methods.

Scientific Applications:

• Systems biology investigations: Improves construction and analysis of gene co-expression networks used in systems biology studies.
• Human disease research: Supports analysis of co-expression patterns relevant to human disease.
• Evolutionary studies: Aids investigation of co-expression relationships pertinent to evolutionary biology.
• Organismal development: Facilitates study of gene co-expression during organismal development.
• Low-sample gene-expression analysis: Enables construction of more reliable co-expression networks from data sets with limited numbers of measurements per gene.

Methodology:

Computes weighted topological overlap by incorporating shared network-neighborhood of gene pairs; optionally applies a soft-threshold modifier to link weights before computing wTO; and compares wTO scores to simple correlation calculations using ensembles of randomly selected measurements (10, 20, 50) against a reference data set with 338 gene-expression measurements per gene.