search

FelixKrueger / Bismark

A tool to map bisulfite converted sequence reads and determine cytosine methylation states
http://felixkrueger.github.io/Bismark/
GNU General Public License v3.0
366 stars 101 forks source link

Correcting low % methylated regions #679

Closed gevro closed 1 week ago

gevro commented 1 week ago

Hi, My % methylation histogram (plotted by methylKit, attached) is not fully bimodal. I'm curious if there is a common practice or would it make sense to manually transform sites with < 10% or < 20% or some other cutoff methylation to 0%? Since one could assume that low % methylated regions are actually unmethylated regions that were not fully bisulfite converted.

Thanks M10.methylStats.pdf

FelixKrueger commented 1 week ago

I don't think this is common practice at all, at least not for bulk data. You need to think about these values over different regions of measurement as the average over several (hundred/thousand/million?) cells. You can say that regions between 20 and 80% of methylation appear more variable (and therefore maybe more interesting?) than regions that are just always methylated or unmethylated.

gevro commented 1 week ago

I see, thanks, makes sense.

gevro commented 1 week ago

Perhaps this explains also why we do not see a more drastically bimodal distribution. Because these are from bull tissues with many different cell types.

FelixKrueger commented 1 week ago

That is indeed most likely the case