mBPCR

mBPCR estimates DNA copy number profiles by identifying regions with copy-number alterations (deletions or gains) using a modified Bayesian Piecewise Constant Regression framework for noisy genomic data.

Key Features:

• Bayesian Piecewise Constant Regression (BPCR): Implements the BPCR framework to model noisy observations as a piecewise constant function and to infer underlying segment structure.
• Segment parameter estimation: Estimates the number of segments, segment endpoints, and segment-level values within the inferred piecewise constant function.
• Modified estimators: Incorporates modifications to the segment number estimator and the boundary estimator to improve breakpoint determination and avoid overestimation of identical-position breakpoints.
• Variance estimation: Provides an alternative method for estimating the variance of segment levels to improve performance on high-noise datasets.
• Performance evaluation: Demonstrates improved accuracy in detecting true breakpoint positions and small aberrations through comparative analyses.

Scientific Applications:

• Cancer genomics: Detection and delineation of somatic copy-number alterations in tumor samples, including deletions and gains.
• Chromosomal aberration analysis: Identification of chromosomal regions with copy-number changes relevant to disease-associated genomic rearrangements.
• Detection in noisy data: Recovery of small copy-number aberrations in highly noisy experimental datasets.
• Therapeutic target discovery: Supporting identification of genomic aberrations that may inform candidate targets for further investigation in clinical research.

Methodology:

Uses Bayesian Piecewise Constant Regression to estimate number of segments, segment endpoints, and segment-level values; applies modified segment number and boundary estimators and an alternative variance estimator; performance assessed via comparative analyses on artificial and real datasets.