Construct RGChannelSet object from raw methylated and raw unmethylated csv files

[giuseppedelnapalle]

Hi minfi developers,

I am trying to analyse a Illumina 450K dataset deposited at GEO. Unfortunately IDAT files are not available. Instead, GSE53051_raw_methylated_signals.csv and GSE53051_raw_unmethylated_signals.csv were uploaded, which are presumably raw signals derived directly from IDAT files. How do I construct RGChannelSet object to utilize functionalities of minfi pacakge?

I found a constructor function RGChannelSet in the manual, but I don't know how to use it. RGChannelSet(Green = new("matrix"), Red = new("matrix"), annotation = "", ...). For type I design, methylated signal does not necessarily correspond to green channel, so it may be incorrect to create the object like rgset <- RGChannelSet(Green = methylated, Red = unmethylated).

Would you give me some advice on this problem?

[kasperdanielhansen]

You can't. You can make a MethylSet at best.

On Tue, Apr 16, 2024 at 5:26 AM Delnero @.***> wrote:

Hi minfi developers,

I am trying to analyse a Illumina 450K dataset deposited at GEO. Unfortunately IDAT files are not available. Instead, GSE53051_raw_methylated_signals.csv and GSE53051_raw_unmethylated_signals.csv were uploaded, which are presumably raw signals derived directly from IDAT files. How do I construct RGChannelSet object to utilize functionalities of minfi pacakge?

I found a constructor function RGChannelSet in the manual, but I don't know how to use it. RGChannelSet(Green = new("matrix"), Red = new("matrix"), annotation = "", ...). For type I design, methylated signal does not necessarily correspond to green channel, so it may be incorrect to create the object like rgset <- RGChannelSet(Green = methylated, Red = unmethylated).

Would you give me some advice on this problem?

— Reply to this email directly, view it on GitHub https://github.com/hansenlab/minfi/issues/269, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABF2DH4I3EKY54B43G2VQE3Y5TVCTAVCNFSM6AAAAABGI6RAB2VHI2DSMVQWIX3LMV43ASLTON2WKOZSGI2DKNJXGA3DKMY . You are receiving this because you are subscribed to this thread.Message ID: @.***>

-- Best, Kasper

[giuseppedelnapalle]

Thanks. I went through the manual and found the MethylSet constructor. I created a MethylSet object using csv files and examined distribution of beta values.

methylated <- read.csv(file.path(dir_i, "GSE53051_raw_methylated_signals.csv"), row.names = 1)
unmethylated <- read.csv(file.path(dir_i, "GSE53051_raw_unmethylated_signals.csv"), row.names = 1)
Mset <- MethylSet(Meth = as.matrix(methylated), Unmeth = as.matrix(unmethylated), annotation = "IlluminaHumanMethylation450k")

Density plot shows the data might not have been normalized. Can I do normalization on MethylSet object? Do you have any suggestions if I do not have access to IDAT files?