“…Given this, we can do a case/control difference analysis for OralUreaseActivityPathways in our independent validation cohort. Here, “case” samples are defined using an aggregate of multiple systemic disease phenotypes which are based on reported associations to urea or uric acid levels in saliva (or blood) [106] , [107] , including: alcohol addiction [108] , [109] , ankylosing spondylitis [110] , diverticulosis [111] , gout [107] , hyperlipidemia [112] , [113] , kidney cyst [114] , [115] , kidney disease [116] , [117] , [118] , pancreatitis [119] , [120] . “Control” samples are selected from those lacking all disease phenotype labels in the validation cohort.…”