Malekpour SA, Pezeshk H, Sadeghi M. PSE-HMM: genome-wide CNV detection from NGS data using an HMM with Position-Specific Emission probabilities.
BMC Bioinformatics 2016;
18:30. [PMID:
27809781 PMCID:
PMC5445519 DOI:
10.1186/s12859-016-1296-y]
[Citation(s) in RCA: 2] [Impact Index Per Article: 0.3] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/22/2016] [Accepted: 10/20/2016] [Indexed: 11/23/2022] Open
Abstract
Background
Copy Number Variation (CNV) is envisaged to be a major source of large structural variations in the human genome. In recent years, many studies apply Next Generation Sequencing (NGS) data for the CNV detection. However, still there is a necessity to invent more accurate computational tools.
Results
In this study, mate pair NGS data are used for the CNV detection in a Hidden Markov Model (HMM). The proposed HMM has position specific emission probabilities, i.e. a Gaussian mixture distribution. Each component in the Gaussian mixture distribution captures a different type of aberration that is observed in the mate pairs, after being mapped to the reference genome. These aberrations may include any increase (decrease) in the insertion size or change in the direction of mate pairs that are mapped to the reference genome. This HMM with Position-Specific Emission probabilities (PSE-HMM) is utilized for the genome-wide detection of deletions and tandem duplications. The performance of PSE-HMM is evaluated on a simulated dataset and also on a real data of a Yoruban HapMap individual, NA18507.
Conclusions
PSE-HMM is effective in taking observation dependencies into account and reaches a high accuracy in detecting genome-wide CNVs. MATLAB programs are available at http://bs.ipm.ir/softwares/PSE-HMM/.
Electronic supplementary material
The online version of this article (doi:10.1186/s12859-016-1296-y) contains supplementary material, which is available to authorized users.
Collapse