Power and false-positive rates for the restricted partition method (RPM) in a large candidate gene data set

BMC Proceedings

Table 1 Power and false-positive rates: all subjects (N = 6479)

CAC	p-value Threshold					R² Threshold					R² Mean^a
	10^-3	10^-4	10^-5	10^-6	10^-7	0.005	0.01	0.02	0.03	0.04
Univariate
Power^b
Het	100	100	100	100	100	100	100	100	29	0	0.0279
ME1	0	0	0	0	0	0	0	0	0	0	0.0001
ME2	85	69	50	32	12	12	0	0	0	0	0.0034
PE1	0	0	0	0	0	0	0	0	0	0	0.0001
PE2	0	0	0	0	0	0	0	0	0	0	0.0001
Avg(#FP)^c	89.8	6.9	3.2	0.9	0.1	3.3	1	1	0.3	0
E(#FP)^d	42.5	4.2	0.4	0	0
Two-way
Power
ME1*ME2	100	100	100	100	100	100	100	100	100	100	0.0539
PE1*PE2	100	100	100	100	100	100	100	100	100	100	0.1095
Avg(#FP)	1.6 M	202 K	28 K	3484	631	68 K	175	<10	<10	<10
E(#FP)	901 K	90 K	9014	901	90
Univariate
Coronary Event
Power
Het	68	42	28	19	11	13	0	0	0	0	0.0026
ME1	0	0	0	0	0	0	0	0	0	0	0.0001
ME2	3	0	0	0	0	0	0	0	0	0	0.0003
PE1	0	0	0	0	0	0	0	0	0	0	0.0001
PE2	0	0	0	0	0	0	0	0	0	0	0.0001
Smoke	98	96	87	71	56	47	1	0	0	0	0.0049
SLoc	0	0	0	0	0	0	0	0	0	0	0.0002
Event Loc	11	2	0	0	0	0	0	0	0	0	0.0008
Avg(#FP)	80.4	5.1	4.6	3.6	2.1	3.2	2	2	1.3	0
E(#FP)	42.5	4.2	0.4	0	0
Two-way
Power
ME1*ME2	100	99	98	86	82	97	29	0	0	0	0.0089
PE1*PE2	100	100	100	100	100	100	100	40	0	0	0.0193
Smoke*SLoc	99	95	87	63	47	77	5	0	0	0	0.0069
Avg(#FP)	1.5 M	173 K	21 K	925	90	96 K	174	<10	<10	<10
E(#FP)	901 K	90 K	9014	901	90

^aMean RPM R² for the factors in the model estimated by the RPM in 100 data replicates
^bPower, number of replicates for which the factor passed the threshold
^cAvg(#FP), average number of observed "false positives" in the univariate analyses per replicate. For two-way analyses, estimated from 10⁸ sampled pairs for the analyses of R² and 10⁷ sampled pairs for the analyses of p-values.
^dE(#FP), expected number of false positives in the analysis of one replicate assuming the tests were independent.

ISSN: 1753-6561