We estimated the proportion of the total phenotypic variance
explained by all SNPs in the combined dataset to be 53.24%. To
determine the phenotypic variance explained by known GWAS
SNPs with the strongest evidence for association with AD and the
two APOE alleles, we controlled for each of these SNPs, and created
an additional dataset with only the APOE alleles. Based on these
analyses, we estimated the phenotypic variance explained by
known GWAS SNPs to be 16%, of which 13% was explained by APOE,
and almost 3% explained by other genes.
A total of 37% of phenotypic variance is tagged by SNPs in our
dataset, but unexplained by known AD SNPs. To determine whether
the unexplained phenotypic variance tagged by genetics is located
adjacent to known AD SNPs or throughout the genome, we created
an additional dataset with all SNPs located in regions of known AD
SNPs (Table 1). We defined a region as 50-kilobases upstream and
downstream of the named GWAS gene, or the gene harboring a rare
variant. We found that 15% and 22% of phenotypic variance tagged
by known disease SNPs is located in regions adjacent to SNPs that
affect risk for AD and outside these regions, respectively. In summary,
of the remaining phenotypic variance that can be explained
by unknown SNPs, approximately 41% is located adjacent to known
AD SNPs and 59% in other genomic regions. Results are summarized
in Table 3.