How to perform multivariate statistical analysis (e.g. PCA) on joint data set of morpho-physio-biochemical analysis (e.g. shoot/root length, fresh weight, chlorophyll concentration, relative water content, MDA content, the activity of various antioxidant enzymes, sodium/potassium ion content etc.) and Proteomics analysis (2DE based protein spots relative abundances) of two different plant genotypes, after the data normalization and Z-score transformation (to provide the data on the similarities between the abundance patterns of the different characteristics studied)?
I suggest you to do two separate PCA for the two data sets, then you will use the extracted components of the two data sets (let's say PCa and PCb) as input for a canonical correlation analysis so to put in light the relation between the two views on a well conditioned frame (principal components are each other orthogonal within their reepsctive data sets so the canonical variates only register between (and not within) data sets relations). More, each single PCA will give rise to a set of loadings more interpretable because refer to the same semantif field (see attached file).
If you want PCA call that package for your data set and see what happens. Alessandro's advise may be helpful for a follow up. Best wishes, David Booth
Thanks for your reply. You are talking about which package? @David Eugene Booth
Dear professor Alessandro Giuliani best for PCA is Matlab
Dear professor Alessandro Giuliani best for PCA is Matlab
Matlab is a good package, I personally do prefer SAS because it is more statistical oriented, byt in any case we are speaking of a technique born in 1901 (see paper attached) when computers did not exist.This means that more or less any sosftware is OK is onlya matter of taste, the important point is to understand the meaning of the statistical tool more than the sostware, exactly like the art of a novel does not depend on the quality of the pen or of the word processor.
Shweta Jha
U can use the R program for multivariate stats.
for PCA- "Mr Bean" for the R platform is available
- Badges
- Science topic