Diablo analysis for differentially expressed omics

Hello everyone,
I’m quite new executing multiple omics integrative analysis (though I do have R and python experience in other fields) hence forgive the newbie question.
I managed to launch the entire DIABLO (n-integration) routine on my data (mRNA from microarray, miRNA and metabolomics from rat brain tissues) as for Chapter 6 Multi-block Discriminant Analysis with DIABLO | mixOmics vignette and everything went quite smooth obtaining a useful list of the selected mRNA, miRNA and metabolites after the long tuning process. Now I was wondering, it is there any way to retrieve the up and down regulated genes/omics from the DIABLO selected list of omics after the block.splsa() and tune.block.splsda() are completed via any mixomics:: package functions?

Basically I would like to recreate the same plots from your work https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6414553/pdf/41467_2019_Article_8794.pdf as for screenshot attached (plus of course retrieving a list of the up and down regulated omics)

I’ve noticed from the same paper (Molecular Interaction Network paragraph at page 11) the overall procedure but It’s still not clear for me whether the differentially expressed genes/omics should be retrieved from a Single Omics analysis (one for mirna, one for mrna and one for metabolites in my case done with limma packages or similar methods) or if they can be obtained from the integrative analysis with Diablo Routine.

Any clarification would be much appreciated.

Many thanks in advance.



Hi @joosefupas,

mixOmics does not perform any differential expression, so if you have come across any such analyses in our publications they most likely would have been from single-omic analyses.

You can however, manually expect the selected variables for up/down regulation using the signs of their loadings. You can use selectVar function to extract these values and segregate the +ve loadings from -ve loadings to see how they are being dysregulated across groups.

Hope it helps,


Dear @aljabadi,
Thanks for your reply and that’s exactly what I’ve done following another useful discussion in this forum!

Best regards