Fig. 6

Internal representation improves outcome prediction compared with the microbiome and when possible also metabolites and is associated with datasets features—on WGS datasets. This figure follows the structure of the previous figure, but for WGS. The datasets used are ERAWIJANTARI, FRANZOSA, MARS, WANG, and YACHIDA for subplot A and for subplot E. In subplots B–H, we used ERAWIJANTARI, FRANZOSA, and the WANG datasets, with the same analysis in Fig. 5B–H. For the categorical variables of B–D, see Supplementary material Table S8. I This subplot follows Fig. 5J just on WGS datasets. The datasets used are ERAWIJANTARI, FRANZOSA, MARS, WANG, and YACHIDA