How to tune (block) sPLS-DA to select all the variable discriminating groups?

Dear mixOmics developers and users,

Is there a particular strategy to fine tune (block) sPLS-DA so that it select all the variables discriminating groups, in the same way a differential analysis would do ?

As far as I recall, sPLS-DA is usually intended to select the minimal set of variables discriminating groups, with the aim to derive a biological signature (and drop those variables that are highly correlated to the ones selected). In my analysis, I would like to be as comprehensive as possible.

Thanks a lot for your advice and tricks.



Dear @Pef
I guess in your case, if you are not interested in variable selection, then you can just use the perf() function on a full block.plsda model to obtain the performance of the model. Note: this is really different from differential analysis, which aims to rank / highlight the features that are DE across conditions. Here you would have a ranking to, if you look at selectVar() on that model.