Methods for discovering genomic loci exhibiting complex patterns of differential methylation

BMC Bioinformatics

Research Article

Thomas J. Hardcastle¹ 

[1] Department of Plant Sciences, University of Cambridge, Downing Street, CB2 3EA, Cambridge, UK;

关键词: Methylation;    DMRs;    High-throughput sequencing;    Epigenomics;    Dicer;   

DOI  :  10.1186/s12859-017-1836-0

 received in 2017-05-19, accepted in 2017-09-11,  发布年份 2017

来源: Springer

【 摘 要 】

BackgroundCytosine methylation is widespread in most eukaryotic genomes and is known to play a substantial role in various regulatory pathways. Unmethylated cytosines may be converted to uracil through the addition of sodium bisulphite, allowing genome-wide quantification of cytosine methylation via high-throughput sequencing. The data thus acquired allows the discovery of methylation ‘loci’; contiguous regions of methylation consistently methylated across biological replicates. The mapping of these loci allows for associations with other genomic factors to be identified, and for analyses of differential methylation to take place.ResultsThe segmentSeqR package is extended to identify methylation loci from high-throughput sequencing data from multiple experimental conditions. A statistical model is then developed that accounts for biological replication and variable rates of non-conversion of cytosines in each sample to compute posterior likelihoods of methylation at each locus within an empirical Bayesian framework. The same model is used as a basis for analysis of differential methylation between multiple experimental conditions with the baySeqR package. We demonstrate the capability of this method to analyse complex data sets in an analysis of data derived from multiple Dicer-like mutants in Arabidopsis. This reveals several novel behaviours at distinct sets of loci in response to loss of one or more of the Dicer-like proteins that indicate an antagonistic relationship between the Dicer-like proteins at at least some methylation loci. Finally, we show in simulation studies that this approach can be significantly more powerful in the detection of differential methylation than many existing methods in data derived from both mammalian and plant systems.ConclusionsThe methods developed here make it possible to analyse high-throughput sequencing of the methylome of any given organism under a diverse set of experimental conditions. The methods are able to identify methylation loci and evaluate the likelihood that a region is truly methylated under any given experimental condition, allowing for downstream analyses that characterise differences between methylated and non-methylated regions of the genome. Futhermore, diverse patterns of differential methylation may also be characterised from these data.

【 授权许可】

CC BY   
© The Author(s) 2017

【 图 表 】

Fig. 9

Fig. 6

Fig. 1

Fig. 2

Fig. 1

12951_2016_171_Article_IEq1.gif

Fig. 1

12936_2023_4742_Article_IEq18.gif