Processing Raw Illumina Data From Geo

Question

[Andre]

Hello,

these are my first steps in processing the data from Illumina (HumanMethylation450). I'm trying to process raw data from http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE37066

require(GEOquery)
getGEOSuppFiles(GEO="GSE37066")

After extracting the raw files, I see that, besides the Chip description there are raw AB intensities. I would like to use lumi or methylumi packages to read the actual data and perform background correction, normalization, etc. to compare the results. However, when executing

require(lumi)
example.lumiMethy <- methylumiR("GSE37066/GSE37066_raw_AB_intensities.txt", sep="\t")

I got an error: "Error in if (dattypes$original[i] != "") { : argument is of length zero" As far as I understand the file GSE37066_raw_AB_intensities.txt is not created by the BeadStudio, right? Is there any method which I can use to convert this data to "MethyLumiSet"? Or is the provided raw data not enough? What do I miss?

Answer 1

[Charles Warden]

The contents of GSE37066_RAW are all annotation files. They are not directly useful for analysis.

It is possible that you could make use of the signal intensities, but I would just use the beta value and detection p-values from primary data matrix:

ftp://ftp.ncbi.nlm.nih.gov/geo/series/GSE37nnn/GSE37066/matrix/GSE37066_series_matrix.txt.gz

I'm not certain about lumi in particular, but you should be able to reformat this matrix to make a data table suitable for analysis. For example, that is how I analyze public 450k data using COHCAP:

http://sourceforge.net/projects/cohcap/

Not sure about your level of bioinformatics experience. The reformatting can definitely be done using a Perl code, and it probably can be done within R as well. If you don't have much coding experience, I know Partek has a GEO downloader application that will make a data matrix for you (even if you don't have an institutional license, you could use a demo license for this feature). Exporting a tab-delimited text file or comma-separated file should allow an easy import into R:

http://www.partek.com/?q=partekgs

[Andre]

Thank you! If I understand this right, the raw data is just incomplete in this case, but the working values are provided with the matrix.

[Charles Warden]

Yeah - GEO won't let you upload the raw .idat files for reasons of space. Therefore, you have to work with some sort of processed data.