Question

MBD-seq non model species

0

Entering edit mode

Marianna ▴ 20

@7cc5052f

Last seen 12 weeks ago

Italy

Hi all,

I'm trying to perform a technical validation of MBD enrichment.

I sequenced a single library (150PE - 1.5 Million reads) as a preliminary test and now I'd like to check if the sequenced reads are on (or close to) CpG islands (i.e. coverage). The target species is Daphnia magna, so the genome is availble, but I guess it is necessary, at first, to "find the CpG islands", and then using sofwares for the coverage assessment. Looking at the most popular software perfroming the methylation analysis (MEDIPS, RaMWAS, QSEA), it seems they do not provide tools for this preliminary assessment. Isn't it?

If someone could give some suggestions, I would be really grateful.

Thank you.

Best

Marianna

MethylSeq ramwas MEDIPS qsea • 1.0k views

ADD COMMENT • link updated 2.3 years ago by Simon Pearce • 0 • written 2.3 years ago by Marianna ▴ 20

0

Entering edit mode

If you have a BSgenome object then reading your data into MEDIPS/QSEA (although I would recommend QSEA) would let you look where the reads fall in relation to the calculated window based CpG density, which should show you a significant skew towards the windows with a higher CG content. But I can't see a BSgenome object for your species.

Another metric is the relative enrichment of the reads compared to the average for your genome, although again this requires a BSgenome in MEDIPS/QSEA. I am working on a package to expand QSEA at the moment. I haven't used RaMWAS so can't comment on that.

ADD REPLY • link 2.3 years ago Simon Pearce • 0

score 0 · Answer 1 · 2022-08-02

Ensembl does CpG island detection "using a program written by G. Micklem, similar to newcpgreport in the EMBOSS package", which links here:

http://emboss.sourceforge.net/apps/cvs/emboss/apps/newcpgreport.html

UCSC CpG island detection is performed using a command line given in this page:

http://genome.ucsc.edu/cgi-bin/hgTrackUi?g=cpgIslandExt

I am not aware of a Bioconductor package to perform CpG island detection, but there might already be a build-in function somewhere? It should be very doable as the task consist in counting C, G, and CpG in a moving window along the genome.