I am analysing a dataset in which we have metabolomics and microbiome data measured on the same samples from a temporal series.
For each dataset I have calculated the Aitchinson distance (CLR transformation + Euclidean distance) between pairs of consecutive samples (e.g. for individual 1: t2 vs t1, t3 vs t2, etc). I would like to compare these distances in the metabolomics vs the microbiome dataset to know which are larger. Would this be statistically correct?
Things that concern me are differences in the number of features and in sparsity between the datasets.
stat | microbiome | metabolome
sparsity | 0.81 | 0.15
nfeatures | 169 | 45
Thanks in advance for any help or pointers on this