KBoost

KBoost infers gene regulatory networks from gene expression datasets to reconstruct regulatory interactions underlying biological processes.

Key Features:

• Methodology: Employs kernel principal components regression (KPCR), boosting techniques, and Bayesian model averaging for network reconstruction.
• Benchmarking: Evaluated against other algorithms using three distinct datasets and reported favorable performance.
• Scalability: Demonstrated capacity to analyze nearly 2,000 breast cancer patient samples involving approximately 24,000 genes in under two hours on standard hardware.
• Prior knowledge integration: Integrates existing knowledge about gene interactions into models to improve prediction reliability.
• Implementation: Provided as an R package.

Scientific Applications:

• Breast cancer subtype analysis: Applied to infer GRNs across molecularly defined breast cancer subtypes, revealing subtype-specific differences in regulatory networks.
• Integration with known interactions: Used to incorporate known gene interaction information into GRN inference to enhance the relevance of predictions.

Methodology:

Kernel principal components regression (KPCR), boosting techniques, and Bayesian model averaging.