PlatinumCNV

PlatinumCNV infers allele-specific copy number genotypes genome-wide from SNP array fluorescent signal intensity data using a Bayesian Gaussian mixture model to analyze allele-specific copy number polymorphisms (CNPs).

Key Features:

• Bayesian Gaussian Mixture Model: Uses a Bayesian GMM to cluster fluorescent signal intensities from SNP arrays into allele-specific copy number genotype clusters.
• Overlap and Error Handling: Allows overlapping cluster memberships to accommodate instrumental errors, absolute scale differences, and X/Y scale imbalances in signal intensities.
• Robustness to Unobserved Genotypes: Remains robust when some allele-specific copy number genotypes are unobserved at a locus.
• Applied to Illumina HumanHap 610K: Has been applied to commercial SNP array intensity data from the Illumina HumanHap 610K platform.
• Discovery of CNPs: Identified over 4,000 allele-specific CNPs, with most showing simple patterns characterized by a single aneuploid haplotype among normal haplotypes.
• Validated Genotyping Accuracy: Achieved mean genotyping error rates as low as 1% validated by quantitative PCR and replicated subjects.
• Foundation for GWAS: Facilitates genome-wide mapping of CNPs to support statistical analyses for disease gene and quantitative trait locus identification.

Scientific Applications:

• Population and genetic diversity studies: Characterize allele-specific copy number polymorphism distributions across samples and populations.
• Disease gene mapping: Map allele-specific CNPs that may be associated with disease genes.
• Quantitative trait locus (QTL) analysis: Identify QTLs linked to allele-specific copy number variation.
• Genome-wide association studies (GWAS): Provide allele-specific CNP genotypes and loci for inclusion in GWAS of complex traits.

Methodology:

Clusters SNP-array fluorescent signal intensities using a Bayesian Gaussian mixture model into clouds corresponding to allele-specific copy number genotypes, allowing overlapping cluster memberships to accommodate absolute scale differences and X/Y scale imbalances and to remain robust to unobserved genotypes.