I’m looking at regression_summary.qzv on my own data generated following the gneiss tutorial here and noticed that the first parameter in each variable (each column in metadata) is missing. For example I have subjects labelled P1, P2, P3 … Pn but P1 is missing. I also have groups C, W, P, and M but C is missing. Is this intended, and if so, why? Thank you very much.
If you are trying to fit a linear model with D categories, you will need to figure out a way to encode that model with D-1 contrasts. As a default, the first category is chosen as the reference contrast. We are using patsy to extract these design matrices, so it may be worthwhile to read the patsy documentation to figure out how to properly encode the formulas.