DMR calling for EWAS meta-analysis

Question

[mark.ziemann]

Hi Biostars,

I'm looking for a package/tool that can perform DMR calling of Infinium array data after meta-analysis. All the packages I've seen so far (eg:DMRcate and bumphunter) require the original matrices of m- or beta-values which won't work for us.

We tried combp from the ENmix package but that gave an error during execution so we are lookig for more options.

We posted the combp error to a different forum https://support.bioconductor.org/p/p132476/

> str(nat_vs_art_combp_filter)

'data.frame':   45151 obs. of  5 variables:

 $ V1: chr  "chr1" "chr1" "chr1" "chr1" ...

 $ V2: int  184943512 39957206 229694784 38273761 95700082 155532716 115053835 27114420 11865920 39457099 ...

 $ V3: int  184943512 39957206 229694784 38273761 95700082 155532716 115053835 27114420 11865920 39457099 ...

 $ V4: num  9.09e-22 3.69e-20 7.06e-20 1.95e-19 4.33e-19 ...

 $ V5: Factor w/ 455828 levels "cg00000029","cg00000108",..: 413831 287383 207196 397045 424258 339026 277278 303634 442510 414728 ...

> combp(nat_vs_art_combp_filter,dist.cutoff=1000,bin.size=310,seed=0.5, region_plot=TRUE,mht_plot=TRUE,nCores=10)

Error in if (any(p.stouffer > 0.05)) { : 

  missing value where TRUE/FALSE needed

Thanks!

[Basti]

Hi, did you find a solution to your issue ? I'm facing the same problem, I would like to use DMRcate method with my meta-analysis output from METAL but I really don't know how to perform this. In this study (https://pubmed.ncbi.nlm.nih.gov/31148503/), they use DMRcate with regression coefficients, standard deviations and uncorrected p-values as input, but I really don't know how they managed to do it. Thanks !

[mark.ziemann]

Hi Basti, no I wasn't able to get DMRcate to work like that. I put together my own function called dmrcaller() to look for runs of coregulated probes using only summary statistics (limma output) here is the link: https://github.com/markziemann/ART_methylation/blob/master/METAL450k.Rmd