How to transform DNA Methylation data in TCGA to effect of methylation on gene expression

Question

[Wayne Lee]

Hello everyone!

DNA methylation data in TCGA describe the methylation extent of single site and corresponding effected genes, but I want to transform it to extent of specific gene effected by all of methylation site. For instance, there have one gene and it could be regulate by 5 methylation sites corresponding to 5 rows in the original data, how to integrate them and calculate a single value to describe this gene's expression effected by methylation? If there have any method or tool to do this job?

Thanks a lot! Wayne

Answer 1

[Kevin Blighe]

Hey Wayne,

What you are aiming to do is... it's an issue in methylatioin studies that is not really discussed or even performed (although I am sure that some Smart Alec reading my answer will go out and trawl the WWW to find some website or 'publication'' where it is very much discussed). The reason is that, unlike gene expression studies where it makes sense to summarise, for example, probe values to gene level, it does not quite make sense when it comes to methylation.

Methylation at different parts of the gene (in genomic co-ordinates) could have very different effects on the resulting transcription of the gene. So, whilst they may each regulate the gene, they do so in different ways. Some of the probes included on the commercial arrays may not even have any baring on gene transcription, or at best minimal.

Even if you wanted to summarise these to gene level, you don't require any 'tool' to do it. Just use a summarisation function, like min, max, mean, or median. In R, you could just use aggregate(). Be aware of the biological interpretation / meaning of doing this, though.

Kevin;.

[Wayne Lee]

Hi, Kevin,

Thanks for your reply!

I really need this information to do some analysis. Can I use the pipeline described on this web page to do this thing? Weather it is better than those summarisation functions?

Thanks!

Wayne

[Kevin Blighe]

Hey Wayne. It seems okay, i.e., the Methylation Preprocessor. I mean, it's from the Broad Institute, which is a trusted organisation that has already worked on TCGA data.

[Wayne Lee]

Vary thanks, Kevin! Your answer helped me a lot.