Is it possible to run DIABLO (or sPLS-DA) on a Y containing a matrix with multiple variables?

Hello everyone.

Thank you for providing us with a great package.
I am now analyzing muti-omics data (transcriptome, proteome, metabolome, etc…) using mixOmics. In our data, DIABLO worked very well to select variables that explain characteristic treatment. For another purpose, I would like to use DIABLO or sPLS-DA to estimate the function of groups of genes (10 genes or more) that have similar expression patterns. Specifically, I would like to investigate what kind of variables in the other omics data (e.g. metabolome and proteome) are related to the characteristic expression pattern of a gene group (Y in block.splsda). I would like to use expression levels of genes as Y in splsda, but I think that Y can only be specified for categorical data in the case of -DA analysis. Which method in mixOmics should I choose for this purpose? Is there no other way but to analyze in regression mode?

Thank you for your advice.

I would recommend giving the below pages a read. They explain the block.spls method. This is the equivalent to block.splsda (i.e. DIABLO) but rather than a categorical response, you can have one (or more) continuous responses.

Read methodology
Read case study

You can swap in different “gene groups” as you please by using a subset of one of your dataframes as the Y input.