[BioC] MEDIPS no longer producing AMS values

[Jonathan Ellis]

Dear Lukas,

Thanks for your response which was very helpful.

Jonathan

On Thu, Aug 15, 2013 at 06:53:42PM -0700, Lukas Chavez wrote:
> Hi Jonathan,
> 
> CpG density normalization of MeDIP-seq data is still possible using MEDIPS
> v >= 1.10.0: by setting the parameter MeDIP of the MEDIPS.meth() function
> to MeDIP=TRUE, the resulting table will contain an additional column for
> CpG density normalized rms values per sample.
> 
> In principle, the concept of CpG density normalization remains the same and
> the dependency between CpG density and read coverage can still be examined
> by plotting the calibration plot. However, there are several conceptional
> modifications in the new version that will result in differences when
> comparing results of the original and of the updated version. For example,
> read coverage and CpG density normalization is performed directly on the
> targeted window size (e.g. genome wide 300bp windows) and not at smaller
> (50bp) bins any more. Consequently, calculation of ams values became
> obsolete. In section 7.7 of the manual, I briefly describe concerns I have
> today with deducing percentage methylation from CpG density normalized rms/
> ams values.
> 
> In summary, i) I recommend to use rms values for comparisons to bisulfite
> data, ii) I consider it to be problematic to deduce percentage methylation
> from the rms values without additional knowledge (see section 7.7 of the
> manual), and iii) I would like to emphasize that the main focus of the new
> MEDIPS package is clearly on the identification of genome wide differential
> methylation (or differential coverage in general) between groups of samples.
> 
> All the best,
> Lukas
> 
> 
> 
> On Thu, Aug 15, 2013 at 6:02 PM, Jonathan Ellis
> <Jonathan.Ellis at qimr.edu.au>wrote:
> 
> > Dear Lukas and Bioconductors,
> >
> > I have been asked to look at a MeDIP-Seq dataset that has been processed
> > with a previous version of the MEDIPS package (version 1.2.0) and I've
> > noticed there have been some changes with regards to the current version
> > I have installed (version 1.10.0).  In particular, it appears as though
> > the whole concept of AMS (absolute methylation score) has been dropped
> > from the most recent version.
> >
> > The previous analysis of the data I'm currently looking at used AMS
> > values, and I hoping someone can tell me why it's no longer used (e.g.,
> > is it no longer considered a useful metric, or is there something else
> > that better describes the methylation?).
> >
> > The 1.2.0 version documentation contains the following (on page 27):
> >
> > "We have shown that such summarized methylation values for ROIs,
> > especially the ams values, are most suitable for comparing MeDIP data to
> > e.g., bisulhpite sequencing or whole genome shotgun bisulphite
> > sequencing data"
> >
> > If this is correct, what is the best metric to use for such comparisons
> > now ams values are no longer being produced?
> >
> > Thanks,
> > Jonathan
> >
> > _______________________________________________
> > Bioconductor mailing list
> > Bioconductor at r-project.org
> > https://stat.ethz.ch/mailman/listinfo/bioconductor
> > Search the archives:
> > http://news.gmane.org/gmane.science.biology.informatics.conductor
> >