Appropriate Pipeline for Methylation EPIC array analysis?

Question

[Ankit]

Hi everyone,

I have two questions.

1. Does anyone knows which tool is the best for Methylation EPIC array analysis? Something as useful as ChAMP.

I know minfi package can help in loading and normalising data. But I also have to filter probes based on SNPs overlap, detection p.value, and so on (similar as all the steps given in champ.load function).

So how should I go about doing it?

I thought to load data with minfi package (not through champ.load) -> filter using champ.filter -> normalize using champ.norm (method =SWAN).

Does it make sense?

2 Also which normalisation method is preferred SWAN or BMIQ? In literature search, I found both as closely equivalent but if I am missing something for data type please correct me.

If any one has prior experience or suggestions please let me know.

Thanks

Answer 1

[mario.red8976]

Hey, I am kinda new to this field but I had to do some analyses on an EPIC array. I followed this R cross-package tutorial for methylation analysis: https://www.bioconductor.org/packages/release/workflows/vignettes/methylationArrayAnalysis/inst/doc/methylationArrayAnalysis.html , I think you will find it helpful.

Regarding your second question, the normalization method depends on what are your starting samples and what you are searching for, probably you can also try different methods and do the analysis to check for differences. I suggest you this paper: https://www.nature.com/articles/bjc2013496 , they do a mini-review on methylation analysis and they also do normalization methods comparison. Hope this helps!

[Ankit]

Hi .

I agree with your suggestions.

But champ package allow lot of filtering at one step command. It also allow several other processing of data.

That's why I thought to load data with minfi and then filter using champ.filter

Answer 2

[BiostarGuardianAngel]

I've had prior experience working with EPIC methylation data. I've tried minfi, limma, methylanalysis, etc. but ChAMP is by far the best!

Notes on installing ChAMP:

1. Close Rstudio and restart a new R session
2. Follow step 2 in ChAMP guide for installation of ChAMP.
3. When asked if you want to update dependency, select the "n" for no option. Very important
4. When package successfully installed, in section 5.2 when starting to input and filter data, restart your computer first. Otherwise you may get an error for memory limit exceeded.
5. For champ.norm, use default BMIQ instead of SWAN for normalizing because you are able to run parallel processing with more cores and can plot the normalization on a graph.

ChAMP guide