Table 2.

Multiple logistic regression for SNPs associated with hair shape variation in 6068 Europeans from QIMR

Marker	Gene	CHR	EA	OR	Lower95	Upper95	P-value	R2	AUC
rs17646946	TCHHL1	1q21.3	A	0.50	0.45	0.56	2.35E-38	0.04	0.58
rs80293268	ERRFI1/SLC45A1	1p36.23	C	0.40	0.30	0.53	3.33E-10	0.05	0.60
Sex (female)				1.43	1.27	1.62	8.71E-09	0.05	0.61
rs506863	FRAS1	4q21.21	C	1.27	1.17	1.38	6.06E-09	0.06	0.62
rs1556547	OFCC1	6p24.3	G	0.81	0.75	0.88	1.85E-07	0.07	0.63
rs74333950	WNT10A	2q35	G	1.35	1.20	1.51	2.68E-07	0.07	0.63
rs310642	PTK6	20q13.33	C	1.56	1.32	1.86	2.76E-07	0.08	0.64
rs11170678	HOXC13	12q13.13	G	0.78	0.70	0.86	6.99E-07	0.08	0.64
rs11203346	PADI3	1p36.13	G	0.79	0.71	0.88	1.53E-05	0.09	0.65
rs11078976	KRTAP	17q21.2	C	0.75	0.66	0.86	2.73E-05	0.09	0.65
rs2847344*	PEX14	1p36.22	G	0.86	0.79	0.93	3.14E-04	0.09	0.65
rs1999874	LINC00708/GATA3	10p14	A	1.15	1.07	1.25	3.91E-04	0.10	0.65
rs2219783	LGR4	11p14.1	G	1.26	1.11	1.44	5.05E-04	0.10	0.65
rs12997742	TGFA	2p13.3	C	0.88	0.81	0.95	1.13E-03	0.10	0.66
rs499697	LCE3E	1q21.3	G	1.13	1.04	1.22	5.54E-03	0.10	0.66

The markers were ordered according to P-values in the multiple logistic regression and hair curliness is dichotomized as non-straight (1) vs. straight (0); Marker, initial analysis includes sex, age, and 16 SNPs associated with hair curliness, 12 from the current study (see Table 1) and 4 from previous studies (EDAR rs3827760, OFCC1 rs1556547, PRSS53 rs11150606, LCE3E rs499697, also see Supplementary Material, Table S6), non-significant SNPs in the final model are not presented; *rs2847344 is used as a replacement for rs6658216, these 2 SNPs are in nearly complete LD (r² > 0.9); R2, Accumulative Nagelkerke pseudo R2 while the current marker is included; AUC, accumulative Area Under the ROC Curve value while the current marker is included.