Table 6.

Allele parameter estimates in European American-dependent smokers

Haplotype namea	MAFb	nc	Modeld	Initial fite	Final fit (i5)f	P ≠ 0g	P ≠ 1h
*1A(51A)	16.6	341	0.74	0.70	0.66	0.04*	0.14
*1H	7.1	146	1	0.45	0.40	0.3	0.07
*9	6.5	134	0.60	0.25	0.22	0.7	0.02*
*12	3.0	62	0	0.20	0.09	1	0.07
*14	4.1	85	1	0.85	0.70	0.1	0.5
*21	0.9	18	0.74	0.75	0.49	0.8	0.6
P-valuei			0.007	0.00072	0.00065

^aHaplotype names follow official recommendations with the addition of *1A(51A) for clarity.

^bPercent minor allele frequency among European American-dependent smokers.

^cTotal number of alleles among European American-dependent smokers.

^dNormalized allele parameters assumed in the predictive model derived from in vivo metabolism measurements in 189 European Americans (6). 1 = full normal activity; 0 = null activity.

^eNormalized allele parameters determined by maximizing the association between CPD and the predicted metabolism metric, varying each parameter individually while maintaining all other parameters according to the original experimentally derived model.

^fNormalized allele parameters determined by maximizing the association between CPD and the predicted metabolism metric following an EM algorithm until it converged (iteration 5, see Materials and Methods).

^gP refers to the statistical likelihood that the allele parameter is not equal to null activity by chance, comparing the fits of the association with CPD using the ‘final fit’ parameters or the same parameters except changing the single allele parameter to ‘0’.

^hP refers to the statistical likelihood that the allele parameter is not equal to full normal activity by chance, comparing the fits of the association with CPD using the ‘final fit’ parameters or the same parameters except changing the single allele parameter to ‘1’.

ⁱOverall P-value for the association between CPD and the predicted metric using each set of parameters.