The association landscape in the AD survival analysis highly resembles that of SPI1 cis-eQTL analysis in myeloid cells (Fig. 1a). We reasoned that the associations of rs1057233 with AD-related phenotypes may be explained by the regulation of SPI1 expression in myeloid cells, and that conditional analysis of the cis-eQTL signal could help us further dissect this complex locus. Therefore, we conducted conditional cis-eQTL analyses in both Cardiogenics datasets as we did above using rs1057233 (the top SNP for AD survival) and rs10838725 (the top SNP for AD risk). In addition, we also examined rs10838698 (a SNP in high LD with rs1057233 that was directly genotyped in the Cardiogenics dataset) and rs1377416, a SNP in high LD with rs10838725 proposed to be a functional variant in an enhancer near SPI1 that is active in human myeloid cells and in the brain of a mouse model of AD19. It should be noted that rs1057233 is a functional variant that has been shown to directly affect SPI1 expression by changing the target sequence and binding of miR-56933. Rs1057233 and rs10838698 remained significantly associated
