In kevinlul/EpigeneLite: Lite version of EpigenCentral

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

library(EpigeneLite)

Before you begin

Download the tarball of IDATS from NCBI GEO and extract them to the same directory (extdata) as the provided CSV sample sheet.

Usually, you would need to construct a sample sheet first, based on the information about the samples in the NCBI GEO Series page. In this example, it has been done for you. You can use the provided extdata/GSE55491/samplesheet.rss-GSE55491.csv as an example to create future sample sheets. For more information, see the documentation for read_idat.

Read Illumina 450k methylation data

grset <- read_idat("../inst/extdata/GSE55491/samplesheet.rss-GSE55491.csv")

Principal component analysis (cursory)

Before we try to find any differentially methylated CpGs, it might be interesting to see if principal component analysis on the methylation values at all of the CpG sites in the assay reveals anything about our data.

pca_plot(grset)

Finding differentially methylated CpGs (sample sheet required)

Recall that earlier in our sample sheet, we labelled each sample in the Sample_Group column as a control or a case (RSS). Now for the real deal, we use minfi's F-test to find differentially methylated positions between our contrasts of statistical significance. This may take some time to complete.

found_dmp_report <- f_test(grset)

Principal component analysis (relevant CpGs only)

Finally, we can do a principal component analysis on just the roughly 1000 identified positions instead of all of them, and see if the pattern is more apparent.