Understanding loading plot

1

Hi,

Firstly, thank you for an excellent package and examples! I have some general questions about interpretation of the loading plot from the sPLS-DA:

  1. The X axis has no label, what are these numbers? Is this the amount of variance in Y explained by each X variable?
  2. The direction of the bars, if a bar is to the left vs to the right what does this mean?
  3. Is there a way similar to the cowplot package to visualise two plots from mixomics together in R?

Apologies if this is explained somewhere, I couldn’t find it in the documentation :slight_smile:

Thanks

1 Like
2

hi @whooh,

More information here:https://mixomicsteam.github.io/Bookdown/plsda.html 4.7.4 Variable selection outputs and also in this paper: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0160169

Kim-Anh

4

Hi, @kimanh.lecao,

Thanks for the documentation, very helpful!

I am confused on one last thing, the direction of the bars in the loading plot. In my loading plot here Staphylococcus is the most important genera, it has the longest bar so it has the highest multivariate regression coefficient in my model. Staphylococcus is also most abundant in the blue group (high). I am still not sure of the meaning of the direction/ +/- sign on the X scale?

Untitled1
Untitled11116×1156 64.3 KB

I am not sure if the figure would make sense with the blue bars going in the other direction?

5

Dear @whooh,
Please have a look at the references in the earlier post, in particular the PLOS ONE paper that explains in the results section how to interpret such plot. The sign corresponds to the sign of the regression coefficients, which are themselves associated to a particular axes of discrimination between classes.

Kim-Anh

1 Like