Question

R, minfi, remove probes from RG channel set (S4 object, DNA methylation array data)

0

Entering edit mode

Qianhui ▴ 10

@qianhui-16128

Last seen 3.6 years ago

Australia

Hi all,

I have a question about removing probes from the RG channel set (S4 object) generated by minfi package. I found the row names for the RG channel set (S4 object) is the probe location IDs on the array instead of probe names started with "cg", so I feel it's not very convenient to filter probes by probe names for this kind of object. Is there a better way to remove or filter probes from this RG channel set (S4 object) by probe names? Thank you so much!

DNA methylation array • 1.3k views

ADD COMMENT • link updated 3.6 years ago by James W. MacDonald 65k • written 3.6 years ago by Qianhui ▴ 10

score 1 · Answer 1 · 2020-10-01

1

Entering edit mode

James W. MacDonald 65k

@james-w-macdonald-5106

Last seen 30 minutes ago

United States

I don't think it's intended to be a simple thing to subset an RGchannelSet, and it is probably not a good idea to do so. What's your use case? Can you not just convert to a MethylSet using preprocessRaw and then subset? Or better yet, do all the processing and then remove things you don't like at the end?

You could certainly remove things, however. There's nothing stopping you from digging around in the Manifest object to get the probe positions and then remove them from the RGchannelSet. But that's a high level move, and you shouldn't expect (and likely won't get) any help if you run into problems.

ADD COMMENT • link 3.6 years ago James W. MacDonald 65k

0

Entering edit mode

Hi James, thank you for your answer! I am doing the functional normalisation for my EPIC array data, and I think it could be better if I filter out the failed probes (including probes with detection p-values > 0.01, bead count <3, cross-hybridisation, and probes with SNPs) before normalisation. However, I found the input object RGchannelSet for the preprocessFunnorm function is not easy to subset because the row names are not probe names. Yes, MethylSet is better than RGchannelSet for probe filtering, I think I could probably modify preprocessFunnorm a bit to use MethylSet as input.

ADD REPLY • link 3.6 years ago Qianhui ▴ 10

0

Entering edit mode

Hi James, thank you for your answer! I am doing the functional normalisation for my EPIC array data, and I think it could be better if I filter out the failed probes (including probes with detection p-values > 0.01, bead count <3, cross-hybridisation, and probes with SNPs) before normalisation. However, I found the input object RGchannelSet for the preprocessFunnorm function is not easy to subset because the row names are not probe names. Yes, MethylSet is better than RGchannelSet for probe filtering, I think I could probably modify preprocessFunnorm a bit to use MethylSet as input.

ADD REPLY • link 3.6 years ago Qianhui ▴ 10

0

Entering edit mode

There's no need to do any of that. Removing those probes isn't likely to have a material effect on the normalization, so you should probably just use the regular pipeline.