The top panel describes the Mann-Whitney test-based taxon disease-control association (TDCA) approach adopted for identifying the alteration patterns of the 201 taxa with a given disease category X (seeSTAR Methods). For each X, each taxon was assigned an association value of SP (significantly positive or disease-associated) or not-significant or SN (significantly negative or health-associated) based on the directionality and the significance. The bottom panel describes the iteration-based approach used for computing the final health-association scores (HS) for all 201 taxa. We performed 10 iterations, each time randomly selecting 18 (of 28) disease categories and applying TDCA on each category. Each taxon was subsequently assigned a score, which was computed as the difference between the fraction of disease categories where it was SN and the fraction where it was SP. This score was ranked across taxa to obtain the iteration-specific scores. Iteration-specific scores were then averaged across 10 iterations to yield the final health-association score (HS) for each taxon.