regioneR: an R/Bioconductor package for the association analysis of genomic regions based on permutation tests

Abstract

Motivation: Statistically assessing the relation between a set of genomic regions and other genomic features is a common challenging task in genomic and epigenomic analyses. Randomization based approaches implicitly take into account the complexity of the genome without the need of assuming an underlying statistical model.Summary: regioneR is an R package that implements a permutation test framework specifically designed to work with genomic regions. In addition to the predefined randomization and evaluation strategies, regioneR is fully customizable allowing the use of custom strategies to adapt it to specific questions. Finally, it also implements a novel function to evaluate the local specificity of the detected association.Availability and implementation: regioneR is an R package released under Artistic-2.0 License. The source code and documents are freely available through Bioconductor (http://www.bioconductor.org/packages/regioneR).Contact: rmalinverni@carrerasresearch.org

MacroH2A histone variants maintain nuclear organization and heterochromatin architecture

“…Repeatmasker (Smit et al) classifies the subfraction of repeat elements that have been annotated so far. We have used the permutation test-based R package regioneR (Gel et al, 2016) to assess whether macroH2A peaks are significantly associated with any class of these repetitive elements. We found that peaks of both macroH2A proteins were positively associated with several types of repeats (Fig.…”

“…A list of all positions shown is given in Table S1. To calculate the association between regions ( peaks or DNA elements) we used regioneR, an R package based on permutation tests (Gel et al, 2016). To compare the associations between them, the calculated z-scores were normalized by dividing them by p n, where n refers to the number of regions in the region sets tested.…”

Epigenetic regulation mechanisms of microRNA expression

“…Among these chromosomes, chromosome 19 stands for which has the highest number of miRNAs overlapping a CpG island (n = 19). Remarkably, using regioneR (48), which assesses the relation between genomic regions using a permutation test, we observed that a higher proportion of miRNA genes than expected by chance is embedded in CpG islands susceptible to methylation (189 of 1881; z-score = 46.13; number of permutations = 1000; p-value = 0.0009, Figure 1B and C, Table 1 and shown in red in Figure 1A), which might explain why certain miRNAs are more prone to be epigenetically regulated by methylation than others. Furuta et al (49) studied the methylation grade of CpG islands close to 39 miRNA genes in hepatocellular carcinoma cell lines and found that miR-124-1, miR-124-3, miR-203 and miR-375, which were completely embedded in a CpG island, underwent methylation-mediated silencing ( Figure 2A).…”

DNA sequence properties that predict susceptibility to epiallelic switching

“…B Venn diagram of genomewide hypomethylated CpG DMRs (chromosomes 1-4) of met1-1 (≥ 80% difference) and combined (union of DMRs) MET1 + plant pools (≥ 40% difference). Permutation test with 1,000 iterations using regioneR package (Gel et al, 2016) confirmed a statistically significant overlap (P-value = 0.001). C Scatter plot of CpG DMRs (chromosomes 1-4) of met1-1 and union of DMRs MET1 + plants (DMR criteria as in A and B).…”