An exploration of internal reference scaling (IRS) normalization in isobaric tagging proteomics experiments. Also, examples of how IRS-normalized data affects statistical testing, and how to avoid using ratios in the analyses.
The IRS method was first described in this publication:
Plubell, D.L., Wilmarth, P.A., Zhao, Y., Fenton, A.M., Minnier, J., Reddy, A.P., Klimek, J., Yang, X., David, L.L. and Pamir, N., 2017. Extended multiplexing of tandem mass tags (TMT) labeling reveals age and high fat diet specific proteome changes in mouse epididymal adipose tissue. Molecular & Cellular Proteomics, 16(5), pp.873-890.
The analysis is of a mouse lens development time course (6 points 3 days apart from E15 to P9) where three replicates of the time points were done in 3 separate TMT labelings. The lens is a unique system that has been studied for many years and the prior knowledge can be used to guide some analysis steps. The data is from this publication:
Khan, S.Y., Ali, M., Kabir, F., Renuse, S., Na, C.H., Talbot, C.C., Hackett, S.F. and Riazuddin, S.A., 2018. Proteome Profiling of Developing Murine Lens Through Mass Spectrometry. Investigative Ophthalmology & Visual Science, 59(1), pp.100-107.
Four jupyter notebook files (R kernel). If you click on the notebook files (*.ipynb extensions), they will render and display in your bowser. Please be patient as they can take a minute to render.
Data from Kahn, et al.:
Sample information for design matrix:
Saved results from the statisticl testing:
Added HTML renderings of the notebooks for those who just want to see the analysis steps and figures (these may load faster):
Added R scripts extracted from the notebooks. These can be used in RStudio or modified for your own analyses.