Gneiss regression issue

Hi
I'd like to perform a correlation analysis with gneiss with qiime2 2018.4. When I use all the covariates (vastedde.tsv) I have the following error: cannot convert float NaN to integer.. then I remove some variable to try, the command works but when I look at the results (regression summary.qzv) something does not work. I attached my files. Please any suggestion will be very appreciated.
greetings

composition.qza (15.3 KB)
hierarchy.qza (17.2 KB)
balances.qza (26.1 KB)
vastedde.tsv (2.2 KB)
regression_summary.qzv (65.2 KB)

Hey there @MarcoT!

You have missing values in a few of your columns (e.g. 2-Heptanol). You will need to filter out these samples so that they aren’t included in this step of the analysis.

Hi Matthew
thanks for your reply. Even if I add some values in the empty cell, I have the same error. Moreover if I run the command including just some covariates it works but the results seem not good. Did you see the regression_summary.qzv?
greetings

@MarcoT, it looks like you are overfitting your data – you only have 8 samples and are fitting with more than 8 covariates.

I’m not sure what you can say with 8 samples. One thing is certain though – if you hope to get something out of the differential abundance, you can only use 1 covariate.

2 Likes

Thanks for your reply

1 Like

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.