Prediction of complex human diseases from pathway-focused candidate markers by joint estimation of marker effects: case of chronic fatigue syndrome

Table 3 Increase in accuracy with increasing number of SNPs in the predictive model with individual-level allelic information

Percentile^a	Cutoff for weighted genetic variation	No. of SNPs	Sensitivity	Specificity	Accuracy
100	3.95	1	74.42	41.38	0.55
95	2.13	9	62.79	70.69	0.67
90	1.30	17	65.12	81.03	0.74
85	1.07	26	67.44	87.93	0.79
80	0.96	34	69.77	87.93	0.80
75	0.82	42	76.74	87.93	0.83
70	0.79	52	81.40	91.38	0.87
65	0.74	58	81.40	89.66	0.86
60	0.70	70	86.05	93.10	0.90
55	0.68	76	86.05	91.38	0.89
50	0.63	84	88.37	96.55	0.93
45	0.58	94	88.37	96.55	0.93
40	0.54	100	88.37	98.28	0.94
35	0.51	109	93.02	96.55	0.95
30	0.48	116	93.02	96.55	0.95
25	0.45	125	93.02	98.28	0.96
20	0.42	135	93.02	96.55	0.95
15	0.41	142	93.02	98.28	0.96
10	0.36	150	93.02	98.28	0.96
5	0.32	159	95.35	98.28	0.97
0	0.00	167	100	100	1.00

^aPercentiles are those for the estimated weighted genetic variation (WGV) under the full model

ISSN: 1479-7364