This is a test version of Biostars. For the public version, visit https://www.biostars.org.
DNA methylation (wig format): Is a quantile normalization across multiple samples neccesary?

Hello everyone,

i have the following problem. I am currently comparing the methylation profile around certain genomic regions. I have downloaded methylomes of three different human tissues/cell types from the GEO database. These methylaomes are all in bed format and look like shown below:

track type=wiggle_0 name="UCSD.Spleen.Bisulfite-Seq.STL003:methRatio" visibility=full ..... very long headline variableStep chrom=chr10
60426 0.8
60427 0.8
60478 0.730769230769231
60479 0.730769230769231 .. ..

As we can see, its just a tab separated format. First column = coordinate of methylated C, Second column = methylation ratio of C.

Since i want to compare the methylation profile around a few selected genomic features across my three samples (cell types), I would like to know:

  1. If a quantile normalization across the samples is necessary.
  2. If it is necessary, is there a good tool or maybe even an R package, that can be used for this task? Unfortunately, these methylomes are very big (~1.3 GB).

Best regards to everybody :)

dna methylation normalization wig
• 2,801 views • link
written by thefirstrealace • thefirstrealace

1 answer

These are standard wiggle files, you can load them with the rtracklayer package.

  1. No, there's no normalization needed.
  2. No applicable

Keep in mind that directly using methylation percentages like this can be problematic. Hopefully the sites all have decent coverage already (of course if this is a public dataset then this may be all that you have...).

• 1 views • link
written by Devon Ryan Devon Ryan

Im glad to hear that i do not have to perform a normalization across my samples, since such a procedure also may lead to a loss of potentially interesting features (from a purely biological point of view). Thanks again for your help.

• 1 views • link
written by thefirstrealace • thefirstrealace

Log in to answer this question.

More posts like this