An integrated Bayesian analysis of LOH and copy number data

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Genomic lesions cause cancer and other disorders.
• SNP-microarrays measure genotype and copy number (CN) at Single Nucleotide Polymorphisms (SNPs).
• Loss of heterozygosity (LOH) is a genomic event that can be better identified by combining CN and LOH data.

Purpose of the Study:

• To propose a method (gBPCR) for inferring genomic aberrations.
• To account for influences on SNP homozygosity detection due to altered CN levels.
• To improve the identification of genomic aberrations by integrating CN and LOH data.

Main Methods:

• Developed gBPCR, a method based on modified Bayesian Piecewise Constant Regression.
• Modeled distributions of detected genotypes given specific genomic alterations.
• Estimated parameters using public reference datasets and improved breakpoint detection estimators.

Main Results:

• gBPCR effectively infers genomic aberrations by considering CN influences on SNP homozygosity.
• The method was evaluated using artificial and real data.
• gBPCR outperformed established methods for LOH estimation.

Conclusions:

• gBPCR enhances the estimation of LOH and CN aberrations by integrating both data types.
• The method effectively accounts for the relationships between CN and LOH data.
• CN lesions estimated by gBPCR on real data were validated using an alternative technique.