Principal Coordinate Analysis PC1, PC2 and PC3

[Niccolo`]

Hi everybody,

I am writing to ask an explanation about the visualization of the results of the beta diversity analysis using Unifrac distances and Principal Coordinate Analysis.

In my case I have 2 animals with 4 different treatments.

When I look at the PCoA plots, I see the PC1 vs PC2, the PC3 vs PC2 and the PC1 vs PC3 plots. My questions are:

1) Does each PC (1, 2 and 3) correspond to one of the coordinates that are used to display my objects?

2) Why are they 3?

3) Which of the 3 graphs should be used for example to display my results in a paper? On which basis to choose?

Thank you very much,

I would really appreciate any help on this

Best regards

Niccolo'

[Sophie]

Hi Niccolo,

There are usually 3 axes displayed because those three capture the majority of the variation in the data.  I would suggest making 3D plots for your paper/analysis - you can do this using make_3d_plots.py or make_emperor.py in 1.8.0 from your unifrac principal coordinates file.

Thanks,

Sophie

[Niccolo`]

Dear Sophie

thanks for your reply

In case I wanted to use the 2D plots, how can I choose the best plot among those three? Simply the one that looks better to explain my question? or considering the variation explained?

Thank you

Best regards

Niccolo'

[Sophie]

Hi Niccolo,

As long as the third axis is above ~7% variance explained, preferable 10% - it is ok to use any of the three plots.  You also want to choose the view that is important to the point you want to make.

Thanks,

Sophie