modSaRa

modSaRa detects copy-number variations (CNVs) and genomic change-points from high-throughput genotyping and SNP intensity data using an enhanced Screening and Ranking algorithm (SaRa).

Key Features:

• Enhanced SaRa: An adapted Screening and Ranking algorithm (SaRa) tailored for CNV detection from SNP and high-throughput genotyping intensity data to address multiple change-point detection challenges.
• Quantile Normalization: Applies quantile normalization to intensity data to stabilize variance across samples and support the normal mean model-based SaRa.
• Normal Mixture Model and Modified BIC: Uses a normal mixture model for candidate change-point selection and a modified Bayesian Information Criterion (BIC) to cluster CNV segments into distinct copy number states.

Scientific Applications:

• CNV Detection: Detecting copy-number variations from SNP array and high-throughput genotyping intensity profiles.
• Change-Point Identification: Identifying multiple genomic change-points (breakpoints) in SNP and genotyping intensity data.
• Copy-Number State Clustering: Segmenting CNV regions and assigning segments to distinct copy number states via mixture modeling and model selection.
• Performance Benchmarking: Comparative evaluation of change-point detection performance against Circular Binary Segmentation (CBS) and validation using HapMap project data.

Methodology:

An enhanced, normal mean model-based SaRa is applied to quantile-normalized intensity data, with candidate change-point selection performed by a normal mixture model and segments clustered using a modified BIC.