Table 1.

Prediction models based on candidate genes.

	T = 1					T=∞
method	Genea	SNP1b	SNP2c	AUC1d	AUC2e	Gene	SNP1	SNP2	AUC1	AUC2
Sig. Thresh.f	130	1176	1145	0.705(0.033)	0.687(0.032)	184	1314	1281	0.730(0.017)	0.715(0.013)
PSg	130	2331	2310	0.645(0.024)	0.627(0.026)	184	3279	3256	0.645(0.024)	0.626(0.026)

^aGene is number of genes identified as GBA associations,

^bSNP1 is number of SNPs mapped to these genes.

^cSNP2 is ^bSNP1 excluding snps that reach GWAS significance of 5 × 10⁻⁷.

^dAUC1, mean AUC of prediction models built with ^bSNP1, numbers in parentheses are standard deviation in cross validations.

^eAUC2, similar to AUC1, prediction models built with ^cSNP2.

^fSig. Thresh. = 0.01

^gPS means genes with highest posterior scores are selected.