I essentially have 6 spectral curves, 3 repeats for two samples. I want to be able to show that two samples are distinct (if they are). I have tried spearman's correlation already, which does produce p values however I feel like I've very much hacked them and that the p value doesn't tell me if they curves are different like an anova test would say. One idea I had was doing PCA and anova on the first PC. Would this be valid?
Any suggestions welcome.