PCoA readouts

[Jeff Galley]

On our data, we have individual samples on a 2D PCoA plot. How exactly
is the program finding these principal coordinates from our distance
matrix? I just don't understand that step, from the matrix to the 2 or
3D plots... I understand the concept behind the PCoA.

[Antonio González Peña]

Hi Jeff,

I do not follow your question, are you asking about the algorithm, the
implementation or both? Anyway, I think a good resource to answer the
first option is this one and also other ordination techniques:
http://ordination.okstate.edu/overview.htm Now if you are interested
in the actual code you need to look in cogent.cluster.metric_scaling
inside pycogent.

Let us know if this doesn't answer your question.

Cheers

--
Antonio González Peña
Research Assistant, Knight Lab
University of Colorado at Boulder
https://chem.colorado.edu/knightgroup/

[antje0402]

Hi Antonio,

I have a related question about the PCoA biplots:
As I can understand from the difference between distance-based and eigenanalysis-based ordination techniques, distance-based methods do not allow for creating biplots (as all information about species identities is hidden once the distance matrix is created). So how does this work in Qiime?

Best,
Antje

[Dan Knights]

Hi Antje,

The biplot coordinates for a given taxon are created by taking a
weighted average of the positions of all samples along a given PC
axis, where the weights are the relative abundances of that taxon in
the samples.

Best,

Dan

[antje0402]

Thanks, Dan!!