Running Qiime2-2019.7, most recent version of miniconda on Mac terminal.
I’m trying to use gneiss to analyze differential abundance of strep bacteria in samples with different phenotypes. However, some of my samples are either controls or environmental samples and I do not want to include them in this analysis. However, leaving the phenotype column blank in gneiss ols-regression analysis leads to an error (“cannot convert float NaN to integer”). My environmental and control samples do not have a phenotype, so how I can either separate out only the samples with a phenotype or bypass the error on ols-regression?