Table 2.
Overview of results of multivariate analysis (PLS-DA) of various data sets.
Original data set | After variable selection | |||
---|---|---|---|---|
# variables | error rate | # variables | error rate | |
plasma proteomics | 66 | 21% | 46 | 10% |
plasma oxylipids | 30 | 26% | 19 | 10.5% |
plasma GC-MS | 137 | 42% | - | - |
plasma LC-MS polar | 130 | 53% | ||
plasma LC-MS lipids | 61 | 37% | - | - |
plasma LC-MS free fatty acids | 14 | 37% | - | - |
PBMC transcriptomics | - | - | 3355 | 0% |
All data sets were analyzed for differences between diclofenac and placebo treatment by PLS-DA. The results of the different multivariate models are expressed as error rates. Only when the error rate for the original data set was < = 30%, analysis was continued by variable selection, as described in the materials and methods section.