Table 1.

Application of DICE to the Association Between Myocardial Infarction and Nine Polymorphisms of the SELP Gene, Using a Codominant Coding Scheme

	Model^a	*AIC_ci*	Δ_i^b	#par^c	Δ_s^d
Step 0	y = intercept + country	1590.07	0.00	2	—
Step 1	T715P	1582.73	0.00	3	7.35
	country^*T715P	1584.16	1.44	4	5.91
	country^*T-1817C	1589.49	6.76	4	0.58
	T741T	1589.98	7.25	3	0.10
Step 2	*T715P^T741T**	1577.34	0.00	5	5.39
	T715P^*N562D	1580.28	2.94	5	2.45
	T715P^*V599L	1580.41	3.07	5	2.32
	country^*N562D	1580.54	3.20	5	2.19
Step 3^e	country^*N562D	1576.27	0.00	7	1.07
	N562D	1576.35	0.08	6	0.99
	T741T^*T-1817C	1577.34	1.06	7	0.00
	T715P^*N562D	1577.42	1.14	7	-0.07
Step 3^bis^f	*S290N ^N562D^g**	1570.84	0.00	8	6.50
	country^S290N^N562D	1572.13	1.29	11	5.21
	*T-1817C^N562D^g**	1572.52	1.68	8	4.82
	T741T^S290N^N562D	1573.39	2.55	11	3.95
Step 4/path1	*S290N^N562D^V599L*	1563.42	0.00	12	7.42
	T-1817C	1571.20	7.78	9	-0.36
	country^*T-1817C	1571.32	7.90	10	-0.48
	N562D^*-485I/D	1571.85	8.42	10	-1.01
Step 4/path2^e	N562D^*A-1969G	1572.80	0.00	10	-0.27
	N562D^*S290N	1573.11	0.31	10	-0.59
	N562D^*-485I/D	1573.50	0.70	10	-0.98
	-485I/D	1573.58	0.78	9	-1.06
Step 4^bis/path2^f	*S290N^N562D^V599L*	1566.15	0.00	14	6.37
	country^C-2123G^S290N	1573.84	7.69	14	-1.32
	N562D^-485I/D+N562D^S290N	1574.07	7.92	12	-1.55
	country^C-2123G+country^S290N	1574.18	8.03	12	-1.65
	Conditional exclusion: deletion of T-1817C^N562D and T-1817C*
	Stop for both paths
*Final model in both paths: y = country + T715P^T741T + S290N^N562D^V599L**

n = 1147. Each model represents an update of the best approximating model identified at the previous step. The four best models within each step are shown. In bold are the best model(s) retained at each step
↵a Symbols are as follows: y is the outcome, “+” means addition of term(s), “^*” means interaction (implying that all the terms of lower order are included in the model according to the principle of marginality)
↵b Δ_i = AIC_ci - minAIC_c
↵c Number of parameters in the model
↵d Δ_s = minAIC_c(s-l) - minAIC_c(s)
↵e Nc model satisfying the composite condition
↵f Combinations of two markers
↵g Tie models

Automated Detection of Informative Combined Effects in Genetic Association Studies of Complex Traits