Table 2.

P values for associations of modules to clinical and pathologic traits.

Module	Activated microglia	Pathologic AD	Clinical AD	Cognitive decline	Neuritic plaques	Diffuse plaques	Neurofi-brillary tangles	Tau	Βeta-amyloid	Sex	Age
M116	0.042 (2.7)	0.16 (0.048)	0.15 (0.049)	0.2 (−0.0094)	0.091 (0.056)	0.61 (0.017)	0.68 (−0.0096)	0.58 (0.048)	0.016 (0.18)	0.013 (−0.082)	0.0049 (1.3)
M115	0.041 (3.1)	0.47 (0.025)	0.16 (0.047)	0.094 (−0.012)	0.099 (0.055)	0.66 (0.015)	0.77 (−0.0068)	0.87 (0.014)	0.022 (0.18)	0.0044 (−0.094)	0.4 (0.39)
M114	0.02 (2.8)	0.0069 (0.11)	0.0045 (0.11)	0.0035 (−0.025)	0.00031 (0.14)	0.055 (0.076)	0.24 (0.033)	0.024 (0.23)	0.00015 (0.34)	0.00016 (−0.15)	0.5 (0.37)
M113	0.63 (0.64)	0.43 (0.028)	0.026 (0.077)	0.085 (−0.013)	0.1 (0.056)	0.28 (-0.037)	0.75 (0.0075)	0.87 (−0.014)	0.0043 (0.22)	0.13 (−0.052)	0.11 (0.75)
M5	0.00042 (4)	0.005 (0.12)	0.068 (0.077)	0.014 (−0.022)	0.0019 (0.13)	0.00036 (0.15)	0.43 (0.023)	0.0028 (0.33)	0.0064 (0.26)	0.00015 (−0.16)	0.42 (0.46)

Linear regression analysis was used to associate average module expression to neuropathologic variables and cognitive decline for all 540 ROSMAP subjects. Reported are the P values with effect sizes in brackets. All the associations were adjusted for age, sex, study (ROS or MAP), RIN, and postmortem interval (PMI). The tests with P values <0.016 were significant after a Benjamini–Hochberg correction for multiple hypothesis testing and are highlighted in bold. Only 105 subjects have activated microglia counts.