Biplot for (s)PLSDA

[AlbanOtt]

Dear mixOmics devs,

Thank you for the package, I enjoy using it.

I am trying to obtain biplot for PLDA sparse or not.

I think the answer here is outdated:
http://mixomics.org/faq/graphical-outputs/
(section : * Can I produce a biplot representation where individuals and variables are superimposed?)
I was not able to use the code, some parameters seems outdated and after few attempts, I can’t reach a plot with both variables and individuals.

Then I read this:

Biplot from PLSDA

Hi @enzo, Thanks for the code. We have in fact implemented a customisable biplot function which is still in development . Below is an example with installation instructions: BiocManager::install('ajabadi/mixOmics@devel', ask=FALSE, update = FALSE) library(mixOmics) data(breast.tumors) X <- breast.tumors$gene.exp colnames(X) <- paste0('GENE_', colnames(X)) rownames(X) <- paste0('SAMPLE_', rownames(X)) Y <- breast.tumors$sample$treatment plsda.breast <- plsda(X, Y, ncomp = 2) bip…

And I was able to reproduce it
Then I tried to apply it to the example from the FAQ but the result seems quite surprising to me.

data(liver.toxicity)
X <- liver.toxicity$gene
Y <- as.factor(liver.toxicity$treatment[, 4])
res = splsda(X, Y, ncomp = 2, keepX = c(20, 20))
biplot(res)

I expected variables to be projected on both component. here they seem to belong to only 1 or 2, never both.

MixOmics : 6.18.0
R 4.1.0

Best,
Alban

[kimanh.lecao]

hi @AlbanOtt ,

This output is expected in the case where you perform variable selection, as the variables selected on Comp 1 will be orthogonal to the variables selected on Comp 2.
If you output a biplot with no variable selection, you will see a big cloud of variables but some will contribute to both Comp 1 and Comp 2.

Kim-Anh