MinimumDistance

MinimumDistance detects de novo copy number variants (CNVs) in case-parent trios from high-dimensional genotyping platforms while reducing technical variation such as probe effects, genomic waves, and batch effects to improve relative copy number analysis.

Key Features:

• Minimum Distance statistic: Uses a univariate "minimum distance" statistic for relative copy number analysis to mitigate technical variation from probe effects and genomic waves.
• Circular Binary Segmentation: Applies circular binary segmentation (CBS) to the minimum distance metric to delineate genomic segments with potential de novo CNVs.
• Maximum A Posteriori Estimation: Performs Maximum A Posteriori (MAP) estimation on segmented data to infer de novo CNV status.
• Computational Efficiency: Provides nearly an 8-fold speed increase compared to the joint hidden Markov model (HMM) implemented in PennCNV.
• Reduction of False Positives: Focusing on the minimum distance statistic reduces false positive CNV calls attributable to genomic waves and batch effects.
• Validation and Concordance: Shows concordance with quantitative PCR validation for CNVs on chromosome 22 in oral cleft case-parent trios, with noted discordance for some high-coverage de novo calls due to genomic waves.

Scientific Applications:

• De novo CNV discovery in trios: Detecting de novo CNVs from array-based genotyping data in case-parent trio designs.
• Genetic studies of oral clefts and complex traits: Investigating the contribution of de novo CNVs to oral clefts and other complex traits.
• Large-scale CNV studies: Enabling high-throughput analyses where computational efficiency relative to methods like PennCNV is advantageous.

Methodology:

Compute the univariate minimum distance statistic on relative copy number data, segment the minimum distance signal using circular binary segmentation, and infer de novo CNVs from segments using Maximum A Posteriori (MAP) estimation; benchmarking versus the joint HMM in PennCNV reports approximately an 8-fold speed improvement.