How to get the explained variance form ordinate

[theo-allnutt-bioinformatics]

As the title says.. using PCoA, I cannot see an explained variance variable or an explanation of the 'trace' variable.

Thanks.

[spholmes]

Hi Theo, It depends which type of ordination you are doing, NMDS for instance doesn't have such a diagnostic and you need to use a stressplot to see how well it is representing the data, on the other hand for a standard MDS the output with contain the eigenvalues which are the explained variances. The example at the end of the this tutorial shows the percentage variance is plotted: https://joey711.github.io/phyloseq/plot_ordination-examples.html

The tricky part is if your data are non Euclidean some of the eigenvalues are negative so the functions usually add the negative one to start with and the computes the explained variance, see for instance: https://stackoverflow.com/questions/54863687/pcoa-function-pcoa-extract-vectors-percentage-of-variance-explained

Hope this helps, Susan

On Tue, Mar 21, 2023 at 5:38 AM Theo Allnutt Bioinformatics < @.***> wrote:

As the title says.. using PCoA, I cannot see an explained variance variable or an explanation of the 'trace' variable.

Thanks.

— Reply to this email directly, view it on GitHub https://github.com/joey711/phyloseq/issues/1668, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAJFZPIHU2TFJF4XEOKHCS3W5E5EBANCNFSM6AAAAAAWB7BOI4 . You are receiving this because you are subscribed to this thread.Message ID: @.***>

-- Susan Holmes

[theo-allnutt-bioinformatics]

Thanks, that is clearer now. Although, still do not know what the 'trace' variable is.

[theo-allnutt-bioinformatics]

Also - I cannot see a plot of the percentage variance in that tutorial.