Quantifying copy number variations using a hidden Markov model with inhomogeneous emission distributions.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Copy number variations (CNVs) are key genetic alterations linked to diseases like cancer and autism.
• High-throughput sequencing generates data for CNV detection, but its statistical properties remain unclear.
• Existing methods for CNV analysis may not fully account for sequencing data biases.

Purpose of the Study:

• To develop a robust statistical model for detecting copy number variations from high-throughput sequencing data.
• To address the distributional properties and sequencing biases inherent in next-generation sequencing data.
• To implement and evaluate a novel algorithm for accurate CNV identification.

Main Methods:

• A hidden Markov model (HMM) was developed with inhomogeneous emission distributions.
• Negative binomial regression was utilized to model sequencing biases within the HMM.
• The proposed model was validated using whole genome sequencing and simulated datasets.
• An R package, CNVfinder, was created to implement the CNV detection algorithm.

Main Results:

• The negative binomial regression-based HMM demonstrated a good fit to sequencing data.
• The CNVfinder algorithm showed competitive performance compared to read count normalization methods.
• The model effectively accounts for biases present in high-throughput sequencing data.
• Accurate detection of copy number variations was achieved.

Conclusions:

• The proposed hidden Markov model provides an effective approach for copy number variation detection.
• Negative binomial regression is a suitable method for modeling biases in sequencing data for CNV analysis.
• The CNVfinder R package offers a competitive and accurate tool for genomic variation studies.
• This work advances the understanding and detection of genetic variations associated with diseases.