Design matrix selection

[Rajanikanth Govindarajulu]

Hello,

I am trying to using mixOmics to integrate transcriptomics and metabolomics data. How would one decide on the appropriate design matrix to input for the block.splsda function. I see there is posting in the FAQ section under the mixOmics site http://mixomics.org/faq/data-input/.

Does that mean i have to run spls function to figure out the correlation structure between the two datasets and use that value in creating a design matrix.

Thanks for any one making suggestions.

-Raj

[Florian Rohart]

Hi Rajanikanth,

This question is discussed in the Diablo paper (on bioarxiv: https://www.biorxiv.org/content/early/2018/03/20/067611).

You can either do pairwise analysis with pls and fill the design matrix with the correlation, or you can also use a full model (by default).

[Al J Abadi]

resolved