Generalized variance

For the generalized variance criteria, a principal component analysis is performed with R software on complete data file.

Generalized variance is calculated from active quantitative and qualitative variables stored into Popdep matrix. The data file has to be complete in order to perform multivariable analysis. First of all, qualitative variables are processed by a correspondence factor analysis.

Statistical calculations are performed with the R software.

The process is :

  1. Ether launching and input files working out to launch R.
  2. These files use active qualitative variables (quali.txt) and active quantitative variables (quanti.txt). Tests are performed on the number of qualitative variables and the number of quantitative variables to determine if multivariable analysis could be done.

  3. R software runs in batch with the input file: quali.txt, quanti.txt and in.txt which contains commands necessary to launch R. Two files are created in ouput:
    the file which gives the number of axes of principal component analysis (axe.txt) and the file giving the coordinates of accessions on these axis (out.txt). The out.txt file is in the vargenr directory. Other temporary files are stored in the vargenr directory and destroyed at the end of the process.

  4. Ether program is launched with the same command line than in the first point. Its goal is checking that quanti.txt, quali.txt and in.txt don't exist anymore and that out.txt does exist. It gives in output construction and redundancy files.

Multivariable analysis have to be performed on complete data file whereas allelic richness optimization can be done on incomplete data file. Complete and incomplete data files will be put into MStrat and be compared on the accession and individual columns. The variables files will have to be same-named and having the same number of variables. Blanks are taken into account in the comparison of the file.


< > ^