cn.mops

cn.mops estimates copy number variations (CNVs) and copy number aberrations (CNAs) from next-generation sequencing (NGS) data by modeling read depth across samples to enable accurate CNV detection for genomic analysis and association studies.

Key Features:

• Data processing: Converts BAM files into read count matrices or genomic ranges objects for per-position coverage analysis.
• Depth-of-coverage modeling: Models depths of coverage across samples at each genomic position to mitigate technological and genomic coverage biases.
• Bayesian mixture modeling: Uses a Bayesian framework with mixture components of Poisson distributions to decompose coverage variation into integer copy numbers and noise.
• Noise estimation and FDR control: Estimates noise levels and filters high-noise detections to reduce the false discovery rate (FDR).

Scientific Applications:

• CNV/CNA detection: Detection and analysis of CNVs and CNAs from NGS datasets.
• Association studies: Association analyses between CNVs and disease or phenotypic traits, with reduced false positives improving statistical power.

Methodology:

Converts BAM files to read count matrices or genomic ranges objects; models per-position coverage across samples; applies a Bayesian mixture of Poisson distributions to infer integer copy numbers and noise; estimates and filters high-noise detections to reduce FDR; benchmarked against five CNV detection methods on four datasets (simulated data; NGS data from a male HapMap individual with implanted CNVs on the X chromosome; HapMap individuals with known CNVs; high-coverage 1000 Genomes Project data) with evaluation by precision (1−FDR) and recall for gains and losses.