Detailed robust and candidate taxon shortlists per dataset.

Each taxon is identified by its internal identifier (‘ID’), as well as its full taxonomy. The complementary information given includes: the taxon’s average Gini importance over all RF models (‘Average_importance’), the list of all annotations associated to the taxon (‘Linked_Reactions’) and the sublist of robustly significant annotations within them (‘Significant_linked_Reactions’), and the profile it is associated to (‘Family’) supported by the average abundances of the taxon in the patient and control samples. Outside of the robust shortlists, the number of SPARTA iterations that deem the taxon significant is also given (‘Count’). (XLSX)