Identifying peaks in *-seq data using shape information

BMC Bioinformatics

Research

Francesco Strino¹  Michael Lappe¹ 

[1] Qiagen Aarhus, Silkeborgvej 2, 8000, Aarhus, DK, Denmark;

关键词: Peak calling;    ChIP-seq;    DNase-seq;    Hotelling observer;    CLC shape-based peak caller;   

DOI  :  10.1186/s12859-016-1042-5

来源: Springer

【 摘 要 】

BackgroundPeak calling is a fundamental step in the analysis of data generated by ChIP-seq or similar techniques to acquire epigenetics information. Current peak callers are often hard to parameterise and may therefore be difficult to use for non-bioinformaticians. In this paper, we present the ChIP-seq analysis tool available in CLC Genomics Workbench and CLC Genomics Server (version 7.5 and up), a user-friendly peak-caller designed to be not specific to a particular *-seq protocol.ResultsWe illustrate the advantages of a shape-based approach and describe the algorithmic principles underlying the implementation. Thanks to the generality of the idea and the fact the algorithm is able to learn the peak shape from the data, the implementation requires only minimal user input, while still being applicable to a range of *-seq protocols. Using independently validated benchmark datasets, we compare our implementation to other state-of-the-art algorithms explicitly designed to analyse ChIP-seq data and provide an evaluation in terms of receiver-operator characteristic (ROC) plots. In order to show the applicability of the method to similar *-seq protocols, we also investigate algorithmic performances on DNase-seq data.ConclusionsThe results show that CLC shape-based peak caller ranks well among popular state-of-the-art peak callers while providing flexibility and ease-of-use.

【 授权许可】

CC BY   
© Strino and Lappe. 2016