SVEM: A structural variant estimation method using multi-mapped reads on breakpoints

Tomohiko Ohtsuki, Naoki Nariai, Kaname Kojima, Takahiro Mimori, Yukuto Sato, Yosuke Kawai, Yumi Yamaguchi-Kabata, Testuo Shibuya, Masao Nagasaki

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Recent development of next generation sequencing (NGS) technologies has led to the identification of structural variants (SVs) of genomic DNA existing in the human population. Several SV detection methods utilizing NGS data have been proposed. However, there are several difficulties in analysis of NGS data, particularly with regard to handling reads from duplicated loci or low-complexity sequences of the human genome. In this paper, we propose SVEM, a novel statistical method to detect SVs with a single nucleotide resolution that can utilize multi-mapped reads on breakpoints. SVEM estimates the amount of reads on breakpoints as parameters and mapping states as latent variables using the expectation maximization algorithm. This framework enables us to handle ambiguous mapping of reads without discarding information for SV detection. SVEM is applied to simulation data and real data, and it achieves better performance than existing methods in terms of precision and recall.

Original languageEnglish
Title of host publicationAlgorithms for Computational Biology - First International Conference, AlCoB 2014, Proceedings
PublisherSpringer-Verlag
Pages208-219
Number of pages12
ISBN (Print)9783319079523
DOIs
Publication statusPublished - 2014 Jan 1
Event1st International Conference on Algorithms for Computational Biology, AlCoB 2014 - Tarragona, Spain
Duration: 2014 Jul 12014 Jul 3

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8542 LNBI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other1st International Conference on Algorithms for Computational Biology, AlCoB 2014
CountrySpain
CityTarragona
Period14/7/114/7/3

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint Dive into the research topics of 'SVEM: A structural variant estimation method using multi-mapped reads on breakpoints'. Together they form a unique fingerprint.

Cite this