Dear community.
I am working with quantitative data from about 5000 proteins measured on the SOMAscan assay, which outputs RFUs. The samples are plasma samples from pregnant women taken from different blood vessels on the maternal and fetal side of the placenta.
The concentration of each sample was NOT adjusted and uniformed across samples before they were loaded on the assay, meaning that the samples could contain different consecration of proteins. The resulting RFUs therefore give a representation of the concentration of the specific proteins in the respective samples. I don’t see this being an initial problem as differences in concentration levels could be interesting itself. However, as a lot of water moves across the placenta (referred to as the water shift) some samples and vessels have a higher/lower concentration of proteins not because of a higher/lower secretion of proteins, but due to gain/loss of water. This makes it difficult to compare protein levels between different vessels, and has resulted in a lot of frustration in my current research group.
As I have a background from RNA-seq, I was wondering if one could do a library normalization (CPM, TMM etc) on these data. Then you would be completely free from differences in concentration across samples. However, I haven’t found any examples in the literature, and I am unsure whether these kinds of normalization methods are applicable on RFU-data. Any thoughts?
Regards, Ina
