HI all,
We have had some HuProt (protein microarrays) arrays done as well as some preliminary analysis done, giving fold changes and p-values for s simple case-control cohort.
We would like to do some more detailed analysis based on further clinical data (such as stratify cases into acute and chronic vs control). I am familiar with R, and have analysed RNAseq data with it before from count tables, using DESeq2, but know this would not be appropriate for the protein array analysis.
With DEseq I can Strat with a "counts table" and "metadata" table and carry out any of the analysis I would like.
I am however not familiar with protein array data and would like to know which format of data I should ask for from the company that is similar to the normalised "counts" data from RNA seq. Essentially protein abundance per patient. And additionally what R package can I use to analyse this protein a abundance data from this point?
Thanks
I don't know about file formats for protein arrays, but my guess is that the data is going to be similar in format to microarray data: a large number of florescence intensity readings in a spatially arrayed format. Thus you will probably be able to analyse the data with limma, although it will probably take some munging to get the data in the right layout.
Thanks, yes it will be more or less indistinguishable from microarray in many ways. Given that, my question is in order to import and use in Limma what should I ask for.My question is more about the terminology of what to ask for.
I do not want to have to go through the normal QC and background subtraction etc for microarray (in this case protein array) as I have little experience with this. So to use in Limma if I ask for a tab eliminated txt file with proteins (features) in rows and patients in columns, should I ask for "normalised" data. "Corrected" data. "Background subtracted data"? All of them? I have played around with making an ExpressionSet and importing it into Limma etc so this is all fine. I just want to know I am inputting the right kind of data. In DeSeq2 (the package I normally use for RNAseq) for example it is important to input un-normalised reads for the statistical model. But I do not think this is the same for Limma?