Table 5. SNPs significantly associated at p<5*10⁻⁸ and imputation r²>0.4 with an ECG phenotype in the discovery imputation association analysis and their validation in an independent sample.

				Discovery cohort					Replication cohort
Gene	rs number	Coded allele	Non coded allele	Imputed CAF	β (se)	Pvalue	Phenotype	Observed CAF	Observed CAF	β (se)	Pvalue
SEL1L	rs61986295	T	C	0.011	1.504 (0.25)	1.65*10−10	PC3	0.012	0.012	0.148 (0.17)	0.4
SP110	rs146598419	C	T	0.012	1.311 (0.21)	3.99*10−10	PC3	0.003	0.001	0.520 (0.65)	0.42
PLA2G2A	rs7796806	A	C	0.04	0.635 (0.11)	1.46*10−9	PC3	0.036	0.036	0.076 (0.12)	0.51
LINC00301	rs117161099	T	C	0.012	1.315 (0.24)	2.40*10−8	PC3	0.006	0.004	-0.644 (0.32)	0.048
PGR	rs76176654	T	C	0.013	1.27 (0.23)	4.47*10−8	PC3	0.010	0.009	-0.141 (0.22)	0.52

Abbreviations: CAF: Coded Allele Frequency, β (se): effect (standard error) associated with the coded allele

No SNP reached statistical significance set at p<5*10⁻⁸ in the discovery cohort when GWAS was applied to ΔQTcF, normalized ΔTpTe, ΔTAmp (n = 489) and notching in women (n = 298). To increase the sensibility of our discovery phase, we applied the GWAS framework to the three principal components variables derived from ΔQTcf, ΔTpTe and ΔTAmp phenotypes. In the discovery samples, the first principal component (PC) derived from the 3 Δvalues explained 65% of the total phenotypic variance (PC1), the second 21% (PC2) and the last one 14% (PC3). Corresponding values in the replication samples were consistent (67%, 19% and 14%, respectively).