CNGPLD: case-control copy-number analysis using Gaussian process latent difference

Area of Science:

• Genomics
• Cancer Biology
• Bioinformatics

Background:

• Somatic copy-number alterations (SCNAs) in cancer genomes are linked to positive selection and driver genes.
• Existing methods lack the ability to identify differential selection pressures across cancer subtypes, sites, or stages.

Purpose of the Study:

• To introduce CNGPLD, a novel computational tool for case-control somatic copy-number analysis.
• To enable the discovery of copy-number aberrations (amplifications or deletions) that differ significantly between cancer groups.

Main Methods:

• Utilizes a Gaussian process statistical framework to model covariance in genomic copy-number data.
• Implements region-level false discovery rate control for robust identification of SCNA regions.
• Performs case-control analysis comparing cancer cases against cancer controls.

Main Results:

• CNGPLD facilitates the identification of differentially amplified or deleted copy-number aberrations.
• The tool accounts for the complex covariance structure of copy-number data along genomic coordinates.
• Enables more precise detection of genomic regions under varying selection pressures.

Conclusions:

• CNGPLD provides an effective approach for identifying genomic loci with differential selection in cancer.
• This tool enhances the analysis of SCNAs for comparative cancer genomics research.
• CNGPLD is available as an R package for widespread use in the scientific community.