PCoA - Negative Eigen Values

[Prasanth S]

Hi all,

im facing some issues during PCoA analysis . When I run the code with principal_coordinates.py  , Im getting following warning

_principal_coordinate_analysis.py:107: RuntimeWarning: The result contains negative eigenvalues. Please compare their magnitude with the magnitude of some of the largest positive eigenvalues. If the negative ones are smaller, it's probably safe to ignore them, but if they are large in magnitude, the results won't be useful. See the Notes section for more details. The smallest eigenvalue is -0.00413316547915 and the largest is 0.540971278734.
  RuntimeWarning

Could someone please help in this? Also I cant able to plot this values using emperor .

Note : After ignoring this warning and proceed to plot the PCoA values. Im getting following warning

RuntimeWarning : mean of empty slice.

[TonyWalters]

The first warning you can safely ignore due to the large differences in magnitude of the values.

For the second warning I'm not so sure as I've never seen it appear in this script. Maybe there is a mismatch between your SampleIDs in your mapping file and those in the OTU table (did you generate them through QIIME or from outside of QIIME? Did you modify anything by hand?). Another possibility is that there is a mismatch between OTU IDs and your tree tip IDs, and it's throwing out data because of this. How did you generate the OTU table and tree files?

Also, apart from this warning, does it still generate the plots?