Table 2.

Summary statistics for the new genome-wide significant SNPs.

Locus	SNP	Chr	A1	A2	Analysis	P value	Freq1	OR	Beta	SE
CAMK2D	rs10029218	4	A	G	SVS-intersect	1.20E−08	0.12	1.27	0.02	0.00
					SVS-rep-EUR	2.46E−02	0.11	1.11	0.10	0.05
					SVS-rep-TRANS	5.98E−03	0.13	1.12	0.11	0.04
SH2B3-BRAP-ALDH2	rs11065979	12	T	C	SVS-CCSp	9.40E−09	0.42	1.13	0.01	0.00
					SVS-rep-EUR	7.62E−03	0.43	1.08	0.08	0.03
					SVS-rep-TRANS	9.29E−03	0.41	1.08	0.07	0.03
PFH20	rs11697087	20	A	G	CES-intersect	3.20E−09	0.09	1.26	0.02	0.00
					CES-rep-EUR	1.55E−02	0.09	1.10	0.10	0.04
					CES-rep-TRANS	4.76E−02	0.09	1.07	0.07	0.03
5:114799266	rs2169955	5	T	C	CES-CCSc	3.90E−08	0.57	0.90	−0.01	0.00
					CES-rep-EUR	1.48E−02	0.56	0.95	−0.05	0.02
					CES-rep-TRANS	2.22E−02	0.56	0.96	−0.04	0.02
GNAO1	rs3790099	16	C	G	CES-CCSp	4.90E−08	0.85	0.87	−0.02	0.00
					CES-rep-EUR	2.97E−04	0.84	0.89	−0.11	0.03
					CES-rep-TRANS	1.10E−02	0.77	0.94	−0.07	0.03

Per locus, the SiGN GWAS is in the first row, in the format ‘subtype–phenotype’. In the other rows, results in MEGASTROKE are shown with ‘subtype-rep-EUR’ for the Europeans-only analysis, and with ‘subtype-rep-TRANS’ for the trans-ancestry analysis. NB, Beta, and SE of SiGN GWAS and MEGASTROKE GWAS are not comparable since they come from linear and logistic regression, respectively. The ORs are comparable. We did a Bonferroni correction: for SVS, α = 0.0125 and for CES, α = 0.00625. Replication p values below the threshold are indicated in bold. Two SNPs (rs2169955:C>T and rs62379973:C>G, in CES-CCSc) that are relatively close (260 kb) on chromosome 5 were in two different clumps, even though they are in LD (r2 = 0.52, D′ = 0.87, in a CEU population [31]) because the distance is just above the threshold (250 kb). Because they are in LD, and just a little farther apart than 250 kb, they were considered to be from the same locus and only the strongest association was kept (rs2169955:C>T).

A1 allele 1, A2 allele 2, Freq1 frequency of allele 1, OR odds ratio, Beta coefficient, SE standard error.