methByRegion: Create plots showing differential methylation and correlation...

In jmacdon/methylation: Package to analyze methylation data from Illumina 450K arrays

Description

A function to create an HTML page with links to plots showing methylation status over each differentially methylated region, a dotplot showing sex-stratified mean methylation, and an HTML table showing correlation between methylation status and gene expression for all genes within 1 Mb of the methylation site. This is the main function for this package, and is likely the only one that an end user should use.

Usage

methByRegion(bmpsObj, eset, samps, contname, longname, txdb, gene.data = NULL,
  chip.db = NULL, fitobj = NULL, fitcol = NULL, cut = 0.001,
  cutcol = c("p.value", "fwer", "p.valueArea", "fwerArea"), dontuse = "",
  orgpkg, stratifyBySex = FALSE, sexfirst = "Male", use.symbols = TRUE,
  extra.indeps = NULL, linearfit = FALSE)

Arguments

bmpsObj	The output from bumphunter
eset	Usually a GenomicRatioSet, created by a call to preprocessQuantile from the minfi package
samps	A data.frame that maps samples to phenotype. This data.frame MUST contain columns named Category and Gender!
contname	A contrast name, used to name the directory where these data will be stored. Usually of the form 'this_vs_that'.
longname	A long form of the contrast name, usually of the form 'This versus that'
txdb	A transcript database (e.g., TxDb.Hsapiens.UCSC.hg19.knownGene)
gene.data	Default is NULL. Otherwise, a data.frame or matrix containing gene expression data. The columns of this data.frame must correspond exactly to columns of the methylation data. Row names must be array IDs or Entrez Gene IDs.
chip.db	The chip-level array data package corresponding to the gene expression data. If NULL, the assumption will be made that the row.names are ENTREZ GENE IDs.
fitobj	An MArrayLM object, created by fitting the same model as used by bumphunter, but probe-wise using the limma package
fitcol	Which column of the MArrayLM object corresponds to the coefficient tested by bumphunter?
cut	The p-value cutoff used to select significant 'bumps'.
cutcol	Which column of the bumpsObj table item to use for defining the p-value cutoff?
dontuse	Which Categories from the samps data.frame should we NOT use? If only two Category levels, use "".
orgpkg	The organism-level annotation package (e.g., Homo.sapiens)
stratifyBySex	Boolean. Should data be plotted stratified by sex? Defaults to FALSE.
sexfirst	Which sex should be plotted first? Defaults to Male. Only used if stratifyBySex is TRUE.
use.symbols	Should transcript IDs be converted to gene symbols when plotting gene regions?
extra.indeps	A character vector of extra independent variables to use when fitting a model regressing gene expression on methylation status (e.g., sex, age, etc). These must conform to column names of the samps data.frame.
linearfit	Boolean. If the underlying model is an ANOVA, this is FALSE, if fitting to a continuous covariate, it is TRUE.

Details

This function is intended to create plots corresponding to a region of the genome that is considered to be differentially methylated, usually after running bumphunter to detect differentially methylated regions. The methylation region plot will consist of the probe-wise methylation, with a smoothed line to indicate the portion that is differentially methylated. The dotplot will show sex-stratified differential methylation (as it is usually safer to do differential methylation for each sex separately). If there are expression data available, there will be links to tables showing the correlation between methylation and gene expression.

Value

This returns an HTMLReportRef that can be used to create an index.html page.

Author(s)

James W. MacDonald ([e-mail])

jmacdon/methylation documentation built on May 19, 2019, 12:49 p.m.