missMethyl vs clusterprofiler Gene ontology analysis

Question

[mikat]

This is my first time dealing with methylation data. I would like to know if my approach is appropriate or biased?

I have a list of differential variable cpgs, and I have to perform Gene ontology analysis. I read multiple published papers and majority of them have used "gometh" function "missMethyl" package. Instead of using "gometh" function, I obtained the Entrez Gene IDs that the significant CpGs are mapped to using "getMappedEntrezIDs" function. Further, performed GO Enrichment Analysis on this gene set using "enrichgo" function of clusterProfiler.

Is this approach correct or is it baised?

Answer 1

[Basti]

I would prefer the gometh approach because it has been specificially designed to adress the biases related to methylation data. Indeed as it is very well explained here (https://genomebiology.biomedcentral.com/articles/10.1186/s13059-021-02388-x), the assumption of independently measured genes is not respected because some CpGs are assigned to several genes. Additionally, some genes are more likely to occur in your DMP list because they inherently contain more CpGs measured by the array. For all these reasons, you should prefer gometh over other methods.

[mikat]

Thank you so much for the explaination.

[mikat]

Hey Basti, can I use the list of genes obtained from "getMappedEntrezIDs" function to link them using PPI networks?

[Basti]

Yes, I do not see any problem with that if you want to interpret your list of genes