BMI-CNV

BMI-CNV integrates whole-exome sequencing and SNP microarray data across multisample datasets to detect copy number variants (CNVs) with improved accuracy for genomic studies.

Key Features:

• Integration of Multiple Genotyping Platforms: Combines whole-exome sequencing (WES) and SNP microarray data to extend genomic coverage beyond protein-coding regions and enable joint analysis of exonic and non-exonic CNV signals.
• Multisample Statistical Modeling: Employs a Bayesian probit stick-breaking process model together with Gaussian Mixture Model estimation to identify shared CNV regions across multiple samples.
• Improved Accuracy and Reduced False Discovery Rate: Extensive simulations reported reduced false positives and increased sensitivity compared to existing methods.

Scientific Applications:

• Accurate CNV Profiling: Supports high-confidence CNV detection for genomic research requiring precise variant calls.
• Large-scale Studies: Applicable to population genetics and disease association studies that integrate diverse datasets.
• Matched Data Analysis (1000 Genomes / HapMap): Detected common variants and expanded the detection spectrum of whole-exome sequencing in matched 1000 Genomes Project and HapMap analyses.
• Cancer Genomics (TRICL): Applied to TRICL data to identify lung cancer risk variant candidates at 17q11.2, 1p36.12, 8q23.1, and 5q22.2.

Methodology:

Integration of WES and SNP array data, Bayesian probit stick-breaking process model to identify shared CNV regions across samples, and Gaussian Mixture Model estimation to estimate CNV distributions.