Cross validation method for OPLS-DA

MetaboAnalyst
, ,
1

Hello everyone!

Do you know where can I verify the default method used for cross validation during OPLS-DA analysis?

Thanks in advance

2

Cross validation is only available for PLS-DA, and OPLS-DA is mainly for its interpretability. We will consider adding support for this feature if necessary. Note PLS-DA and OPLS-DA have Identical Predictive Power and the total predictive variance explained remains exactly the same. Please refer to this paper.

3

Dear Xia,

Thank you for your kind answer and for sharing this paper — it has been very useful.

I apologize for pushing further on this point, but I am still wondering what is the default cross-validation scheme that MetaboAnalyst uses to compute Q² during OPLS-DA analysis.

For example, is the reported Q² based on a specific procedure such as k-fold cross-validation, leave-one-out cross-validation, or another internal validation approach?

If there are supporting documents, methodological references, or guidance within the MetaboAnalyst platform where this implementation detail is described, I would greatly appreciate it if you could point me toward them.

Thank you again for your time and help

4

The technical detail is documented at the top of the OPLS-DA web page

Orthogonal PLS-DA

The implementation of this method is based on the R package ropls as described by this paper.

5

Thank you! I’ve looked onto the ropls documentation and it seems that the default cross validation method is 7 fold cross validation.

I appreciate your help and guidance :smiling_face: