Fig 3. Spurious association of taxa with case-control status without appropriate correction.

(A) We tested the number of associations identified after replacing the controls from the CRC-WGS study sequenced by [1] referred to as Thomas et al. 2018a with controls from Feng et al. at increasing proportions and vice versa. (B) Similarly, controls in the CRC-WGS study Hannigan et al. [79] were replaced with controls from Zeller et al.[74] and vice versa (S7 Fig). BMC + CLR was an outlier and excluded for clarity of visualization, but the summary of mean associations of BMC + CLR is in S1 Table.