I am running through my analysis of a genome which largely has been suggested to have quite high levels of non-cpg methylation, however, having looked through the MEDIPS package I cannot seem to find a way in which I can distinguish between CpG and non-CpG. This may just be that I am a bit novice and am not looking in the right places, but, for example in sequence pattern coverage analysis, the pattern CG is used in the manual....what is the pattern needed if I want to look at all methylated cytosines and their coverage here?
as Lukas pointed out, MeDIP seq (or any other mC enrichment based methylation assay) is not well suited to distinguish between CpG and CpN methylation. However, you can use MEDIPS to judge whether there is a substantial fraction of non CpG methylation by looking at the calibration plots: Following Lukas suggestion, there should be no enrichment in low CpG density regions if methylation is present only in CpG context. This leads to the dependency of mean coverage on CpG density, as can be observed in the calibration plot, for a typical example, see Figure 1D from the MEDIPS update application note: "MEDIPS: genome-wide differential coverage analysis of sequencing data derived from DNA enrichment experiments".
In contrast, in case of substantial non CpG methylation you will observe a somewhat higher coverage at low CpG density, and the slope will not be that steep.
Dear Marc,
MEDIPS does not allow to distinguish CpG and non-CpG methylation. This is because the MeDIP-seq protocol does not allow to distinguish between CpG and non-CpG methylation. You might be able to estimate non-CpG methylation based on MeDIP-seq read enrichment in regions free of CpGs (you could check the coupling factor column of the result table returned by MEDIPS.meth() to find genomic windows with no CpGs (CF=0)). However, whenever a DNA fragment contains both, CpGs and non-CpGs, it is not possible to conclude that the fragment was immunoprecipitated due to methylation at the CpG or at the non-CpG. There might be some clever ways to de-convolute non-CpG from CpG methylation based on MeDIP-seq data, but I am not aware of such an method? If you want to distinguish CpG from non-CpG methylation you probably have to do bisulphite sequencing.
The MEDIPS.seqCoverage() offers the parameter pattern which is set to “CG” by default. It should be possible, in principle, to change this to C to calculate the coverage at all Cs. However, the amount of C’s in the reference genome is typically >>> than the amount of CpGs, so I do not know, how efficient the function will run, I have never tested it.
All the best,
Lukas
> On 18 Mar 2015, at 13:08, mamd201 [bioc] <noreply@bioconductor.org> wrote:
>
> Activity on a post you are following on support.bioconductor.org <https: support.bioconductor.org=""/>
> User mamd201 <https: support.bioconductor.org="" u="" 7479=""/> wrote Question: Non-CpG Methylation in MEDIPS Package <https: support.bioconductor.org="" p="" 65781=""/>:
>
>
> Dear All,
>
> I am running through my analysis of a genome which largely has been suggested to have quite high levels of non-cpg methylation, however, having looked through the MEDIPS package I cannot seem to find a way in which I can distinguish between CpG and non-CpG. This may just be that I am a bit novice and am not looking in the right places, but, for example in sequence pattern coverage analysis, the pattern CG is used in the manual....what is the pattern needed if I want to look at all methylated cytosines and their coverage here?
>
> Any help would be appreciated.
>
>
> Many thanks,
>
>
> Matt
>
>
> You may reply via email or visit Non-CpG Methylation in MEDIPS Package
>
