Filtering DMRs

Question

[niutster]

Hi,

I get a large number of DMRs by seqlm package in comparison between two tissues after filtering by p-value <0.05 , it is near 115000 regions which contains one bp differentially methylated sites,too. Does anyone of you have recommendations for filtering these regions to get some meaningful sites and regions? I have features like number of probes have changed, CpG islands, different regions in genome such as body, UTR,... and beta value.

[mark.ziemann]

Are you doing WGBS, array, MBD or other profiling protocol? That's a lot of DMRs. Try using GREAT (http://bejerano.stanford.edu/great/public/html/) to annotate DMRs with gene names. GREAT also does a reasonable job at GO enrichment. Alternatively, you can annotate DMRs with gene names using a BedTools with an bed file of gene/TSS coordinates. Once annotated with gene names, a pathway analysis using GSEA would be insightful.