Bayesian Copy Number Variation Analysis
Also known as: Bayesian CNV analysis, Bayesian CNV calling, probabilistic CNV detection, Bayesian HMM-CNV
Bayesian copy number variation (CNV) analysis is a probabilistic framework for detecting genomic segments where an individual's DNA copy count deviates from the diploid norm. By placing prior distributions over copy-number states and updating them with array CGH, SNP array, or sequencing read-depth evidence, the approach yields posterior probabilities for each copy-number state along the genome, providing statistically principled uncertainty quantification that frequentist segmentation methods lack.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use Bayesian CNV analysis when you need statistically principled, uncertainty-aware copy-number calls from array CGH, SNP arrays, or whole-genome sequencing data — especially in clinical genomics, rare-disease gene discovery, or cancer genomics where false positives are costly. It is particularly valuable when sample sizes are small and confidence intervals on CNV calls matter. Do not use it as a simple drop-in replacement for standard pipelines if computational resources are severely limited, as Bayesian HMM fitting is more expensive than circular binary segmentation. Avoid applying it when the input data quality is very low (poor coverage, high batch effects not correctable by normalization), since even a well-specified prior cannot rescue severely corrupted signal.
Strengths & limitations
- Provides posterior probabilities for each copy-number state, enabling principled uncertainty quantification and confidence-ranked CNV calls.
- The HMM structure exploits spatial autocorrelation along the genome, reducing false positives from isolated noisy probes.
- Prior distributions can encode biological knowledge about CNV frequency and length, improving detection in low-coverage or small-sample settings.
- Objective Bayes formulations (e.g., QuantiSNP) remove the need for subjective prior elicitation while retaining probabilistic calibration.
- Outputs are directly interpretable as probabilities, facilitating downstream Bayesian meta-analyses and multi-sample joint calling.
- Computationally more demanding than frequentist segmentation algorithms (e.g., CBS); scales poorly to thousands of samples without parallelization.
- Model misspecification — e.g., choosing too few copy-number states or an inappropriate emission distribution — can produce systematically biased calls.
- Prior choice influences results, particularly in sparse data regions; non-informative priors may underperform informative priors when biological knowledge is available.
- Standard HMM-based implementations assume stationarity and may not handle complex structural variants (inversions, translocations) or highly aneuploid cancer genomes well.
Frequently asked
How is Bayesian CNV analysis different from circular binary segmentation (CBS)?
CBS is a frequentist change-point method that partitions the genome into segments of equal mean signal using permutation tests; it produces point estimates of segment means without uncertainty quantification. Bayesian CNV analysis places probability distributions over copy-number states and updates them with data, yielding posterior probabilities for each state at every position. This means Bayesian methods naturally express confidence in each call, handle noisy data more gracefully, and integrate prior biological knowledge — advantages that come at the cost of greater computational complexity.
Which input data types are supported?
The framework applies to any signal that reflects DNA dosage: log R ratios and B-allele frequencies from SNP arrays, probe-level log2 ratios from array CGH platforms, and read-depth profiles from whole-genome or whole-exome sequencing. Each data type requires tailored normalization and a suitable emission distribution within the HMM (e.g., Gaussian for log-ratio data, beta-binomial for allele frequencies).
Can this method be applied to single-cell sequencing data?
Yes, but with important modifications. Single-cell whole-genome sequencing data is typically very sparse (low coverage per cell), so the prior on segment length and copy-number states plays a larger role. Tools such as HMMcopy and SCOPE adapt Bayesian HMM frameworks to the single-cell setting by pooling information across cells or leveraging shared priors, enabling inference of clonal CNV structure.
How many samples are needed for reliable calls?
A single sample can yield CNV calls, which is one advantage over methods that require reference panels. However, batch correction and normalization improve substantially with more samples in the same experiment. For population-level CNV discovery, larger cohorts (hundreds to thousands of samples) are needed to distinguish rare variants from technical artifacts and to estimate allele frequencies accurately.
What software implements Bayesian CNV analysis?
Well-known implementations include QuantiSNP (Objective Bayes HMM for SNP arrays), PennCNV (HMM-based for Illumina SNP arrays), HMMcopy (for sequencing read-depth data), BayesCNV, and SCOPE (single-cell). Each tool differs in the choice of prior, emission model, and input data format; selection should be guided by the platform and study design.
Sources
- Colella, S., Yau, C., Taylor, J. M., Mirza, G., Butler, H., Clouston, P., Bassett, A. S., Seller, A., Holmes, C. C., & Ragoussis, J. (2007). QuantiSNP: an Objective Bayes Hidden-Markov Model to detect and accurately map copy number variation using SNP genotyping data. Nucleic Acids Research, 35(6), 2013–2025. DOI: 10.1093/nar/gkm076 ↗
- Fridlyand, J., Snijders, A. M., Pinkel, D., Albertson, D. G., & Jain, A. N. (2004). Hidden Markov models approach to the analysis of array CGH data. Journal of Multivariate Analysis, 90(1), 132–153. DOI: 10.1016/j.jmva.2004.02.008 ↗
How to cite this page
ScholarGate. (2026, June 3). Bayesian Copy Number Variation Analysis. ScholarGate. https://scholargate.app/en/bioinformatics/bayesian-copy-number-variation-analysis
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Bayesian GWASBioinformatics↔ compare
- Bayesian Variant CallingBioinformatics↔ compare
- Copy Number Variation AnalysisBioinformatics↔ compare
- Genome-wide association studyBioinformatics↔ compare
- Single-cell Copy Number Variation AnalysisBioinformatics↔ compare
- Variant CallingBioinformatics↔ compare