DMR {sesame}R Documentation

Find Differentially Methylated Region (DMR)

Description

This subroutine uses Euclidean distance to group CpGs and then combine p-values for each segment. The function performs DML test first if cf is NULL. It groups the probe testing results into differential methylated regions in a coefficient table with additional columns designating the segment ID and statistical significance (P-value) testing the segment.

Usage

DMR(
  betas,
  smry,
  contrast,
  platform = NULL,
  refversion = NULL,
  dist.cutoff = NULL,
  seg.per.locus = 0.5
)

Arguments

betas

beta values for distance calculation

smry

DML

contrast

the pair-wise comparison or contrast check colnames(attr(smry, "model.matrix")) if uncertain

platform

EPIC, HM450, MM285, ...

refversion

hg38, hg19, mm10, ...

dist.cutoff

distance cutoff (default to use dist.cutoff.quantile)

seg.per.locus

number of segments per locus higher value leads to more segments

Value

coefficient table with segment ID and segment P-value each row is a locus, multiple loci may share a segment ID if they are merged to the same segment. Records are ordered by Seg_Est.

Examples


sesameDataCache("HM450") # in case not done yet

data <- sesameDataGet('HM450.76.TCGA.matched')
smry <- DML(data$betas[1:1000,], ~type, meta=data$sampleInfo)
colnames(attr(smry, "model.matrix")) # pick a contrast from here
merged_segs = DMR(data$betas[1:1000,], smry, "typeTumour")

# release memory for Windows package builder
rm(list=ls(env=sesameData:::cacheEnv), envir=sesameData:::cacheEnv)
gc()


[Package sesame version 1.12.8 Index]