I have a list of regions identified as differentially methylated regions using bump hunter in minfi package with the format

chr  start  end ….

chr1

chr2

chr3

….

I want to identify genomic features of these positions (5' UTR, exon, gene body or 3' UTR)

I fought of using findOverlaps and converting txdb to GFF

I converted txdb to GFF

txdbGFF<-asGFF(txdb)

the information about type is gene,mRNA,exon,CDS 

unique(txdbGFF$type)

[1] "gene" "mRNA" "exon" "CDS" 

How can I get more detailed info. like 5' UTR, intergenic,..etc.

Thank you

I would recommend putting your 'list of regions' in a GRanges or GRangesList class if they aren't already.

You don't need to convert the TxDb to a GFF. Instead use the functions in GenomicFeatures to extract the regions of interest, then call findOverlaps().See ?transcripts and ?transcriptsBy for a listing of functions that extract regions from a TxDb.

Another option is to use locateVariants() in the VariantAnnotaiton package. That function performs findOverlaps() between your ranges and a TxDb based on the 'regions' argument you provide. In the example below, 'gr' is a GRanges of your positions of interest.

library(TxDb.Hsapiens.UCSC.hg19.knownGene)

txdb <- TxDb.Hsapiens.UCSC.hg19.knownGene 

This call identifies ranges in 'gr' that fall in coding regions:

loc_all <- locateVariants(gr,  txdb, CodingVariants())

This identifies ranges in 'gr' that fall in 5 UTR regions:

loc_all <- locateVariants(gr,  txdb, FiveUTRVariants())

See ?locateVariants for more 'region' options.

Valerie

You may want to look at the annotatePeak function in the ChIPseeker package that does this.

Vince