Hello @windsnowflake,
If you would like to look at all your variables at once:
- metabolites (232 continuous variables)
- before/after intervention (1 categorical variable)
- age (1 continuous variable)
- sex (1 categorical variable)
I would first break up the problem to see which of the covariates (before/after, age, sex) strongly affect your metabolite data. You can do this using a PCA plot on your metabolite data and colouring the samples for these different covariates - you might find that some of these covariates do not strongly impact your data and therefore may not be important to include downstream.
It is possible to construct a PLS where your X block is metabolite data and Y block are your other three variables, in this case you turn your categorical variables (before/after intervention and sex) into numeric (1 or 2). See this post for an example of someone doing this with clinical data. The problem is now you don’t have a discriminate analysis model anymore (you would be running PLS rather than PLS-DA), which I would think would be much more difficult to interpret, particularly if you are primarily interested on the effect of intervention on your metabolites.
Instead of building a PLS model, I would recommend trying to correct for your covariates i.e. age and sex before building a PLS-DA model, currently mixOmics does not include this functionality but you can have a look through the forum to see what tools others have used (perhaps this post is useful where someone is correcting for sex).
Hope that helps!
Eva