一篇综述 Analysing and interpreting DNA methylation data

pipeline：

整理自：http://bioconductor.org/help/course-materials/2015/BioC2015/methylation450k.html

source("http://www.bioconductor.org/biocLite.R")

biocLite(c("IlluminaHumanMethylation450kmanifest","IlluminaHumanMethylation450kanno.ilmn12.hg19","minfiData", "sva"，" minfi"))nn

library(minfi)library(minfiData)library(sva)Reading DatabaseDir <- system.file("extdata", package="minfiData")targets <- read.450k.sheet(baseDir)RGSet <- read.450k.exp(targets = targets)Preprocessing and normalizationGRSET <- preprocessFunnorm(RGSet)[preprocessFunnorm] Background and dye bias correction with noob [preprocessNoob] Using sample number 2 as reference level...[preprocessFunnorm] Mapping to genome[preprocessFunnorm] Quantile extraction[preprocessFunnorm] Normalizationannotation <- getAnnotation(GRset)names(annotation)GRset <- dropLociWithSnps(GRset, snps=c("SBE","CpG"), maf=0)Identifying DMRs and DMPsbeta <- getBeta(GRSET)age  <- pData(GRSET)$ageposition analysis:dmp <- dmpFinder(beta, pheno = age  , type = "continuous")head(dmp) DMR analysis:

1.  Define yourphenotype of interest

pheno <- pData(GRSET)$statusdesignMatrix <- model.matrix(~ pheno)

2.  Run the algorithmwith B=0 permutation on the Beta-values, with a medium difference cutoff, say0.2 (which corresponds to 20% difference on the Beta-values):

dmrs <- bumphunter(GRSET, design = designMatrix,              cutoff = 0.2, B=0, type="Beta")

3.  If the number ofcandidate bumps is large, say >30000, increase the cutoff to reduce thenumber of candidate bumps. The rationale behind this is that the most of theadditional candidate regions found by lowering the cutoff will be found to benon-significant after the permutation scheme, and therefore time can be savedby being more stringent on the cutoff (high cutoff).

4.  Once you havedecided on the cutoff, run the algorithm with a large number of permutations,say B=1000:

dmrs <- bumphunter(GRSET, design = designMatrix,             cutoff = 0.2, B=1000, type="Beta")

5.  

转载本文请联系原作者获取授权，同时请注明本文来自张洪磊科学网博客。
链接地址：https://blog.sciencenet.cn/blog-2609994-984986.html

上一篇：生存分析及HR计算
下一篇：STAR 比对软件用法举例