Fig 1.
Population structure analysis reveals relative homogeneity of the discovery population (tan plus sign) compared to the validation population (black x) (p<2.2x10-16).
Shown are the first two principal components calculated from all variants genotyped in both the discovery and validation cohorts. Representative populations from the 1000 Genomes Project: Han Chinese in Beijing (CHB, n = 97; red square), Utah residents with Northern and Western European ancestry from the CEPH collection (CEU, n = 86; blue circle), and Yoruba in Ibadan, Nigeria (YRI, n = 88; black triangle) are plotted for context. The discovery population overlapped with the CHB population (mean distance to CHB = 0.001, CEU = 0.07, YRI = 0.07).
Fig 2.
Top associations between genetic variation and odor intensity perception.
Each row represents the top SNPs associated with the odor intensity phenotype in the discovery cohort (blue filled circles) and the replication cohort (red open circles). Listed next to each odor is the nearest gene to the top SNP for each cohort. There were two novel associations that reached genome-wide significance in the discovery cohort (p < 2.5x10-9, solid blue line): Galaxolide/rs1453541(M263T) and rs1453542 (S151T) (p<2.4x10-25, p<3.0x10-25) and 3M2H/rs3898917 (p<1.9x10-11). The discovery study replicated the associations from the literature (p<0.05, dotted red line) for β-ionone/rs6591536 (D183N) (p<5.5x10-42), androstenone/rs61729907 (R88W) and rs5020278 (T133M) (in both cases, p<9.3x10-8), and cis-3-hexen-1-ol/rs28757581 (T113A) (p<0.02). The discovery cohort was unable to impute the region around OR1A1 for replication of the caproic acid association. Other than β-ionone, no replication odors had associations that reached genome-wide significance. For this set of replication odors (β-ionone, androstenone, cis-3-hexen-1-ol, and caproic acid), the association shown is the top association from the LD-band surrounding the previously identified SNP. In the validation cohort (open red circles), we tested associations for the significant SNPs and surrounding LD-bands (±200kb) from the discovery study and previous literature. There were four significant associations in the validation study (p<0.05; red dotted line): β-ionone/rs6591536 (D183N) (p < 7.8x10-39), Galaxolide/rs591536 and rs7941591 (p<5.8x10-6), 3M2H/rs10837814 (L143F) (p<9.6x10-8), and androstenone/rs61732668(P79L) (p<3.5x10-4). The association for caproic acid/rs17762735 (p<6.9x10-3) is significant, but in the opposite direction predicted by the previous study.
Fig 3.
Galaxolide perception is associated with variation in chromosome band 11q12.1 in both cohorts, as shown by A) a Manhattan Plot of associations with the discovery study in blue and validation study in red (high concentration) and yellow (low concentration). The red line indicates the threshold for genome-wide significance (p<5x10-8). Inset: QQ plots from the discovery (blue) and validation ([high] = red, [low] = yellow) cohorts (Genomic Lambda: discovery = 1.02; validation = 0.90) show appropriate control for inflation due to population structure. B) The regional plot of discovery study associations indicates both the significance level and the recombination rate at the OR4D6 LD-band. Genetic variation in OR4D6 affects the perceived intensity of Galaxolide in C) the discovery cohort and D) the validation cohort (high concentration). The x-axis is ordered left-to-right with increasing number of variant alleles for the M263T variant, for population frequency of M263T indicated below the genotype. The points colored by the S151T genotype suggest that in the validation cohort, S151T is driving the Galaxolide anosmia phenotype exhibited by those homozygous for the variant (T/T).
Fig 4.
3M2H perception is associated with variation in chromosome band 11p15.4 in both cohorts, as shown by A) a Manhattan Plot of associations with the discovery study in blue and validation study in red (high concentration) and yellow (low concentration). The red line indicates the threshold for genome-wide significance (p<5x10-8). Inset: QQ plots from the discovery (blue) and validation ([high] = red, [low] = yellow) cohorts (Genomic Lambda: discovery study = 1.01; validation study = 0.90) show appropriate control for inflation due to population structure. B) Regional plot of discovery study associations indicating both the significance level and the recombination rate at the OR51B2/4 LD-band. The variant L134F (rs10837814; OR51B2) was associated with the perceived intensity of 3M2H in the C) discovery and D) validation (low concentration) cohorts. The x-axis is ordered left-to-right for increasing number of variant alleles, with population frequency indicated below the genotype. E) The OR51B2 reference haplotype responds to 3M2H in a cell-based assay, but the haplotype containing the L134F variant does not. The empty vector control (Rho) does not respond to 3M2H, nor do other receptors in the same LD-band such as OR51B4.
Fig 5.
β-ionone perception is associated with variation in rs6591536 (OR5A1) in both cohorts, as shown by A) a Manhattan Plot of associations with the discovery study in blue and validation study in red (high concentration) and yellow (low concentration). The red line indicates the threshold for genome-wide significance (p<5x10-8). Inset: QQ plots from the discovery (blue) and validation ([high] = red, [low] = yellow) cohorts (Genomic Lambda: discovery study = 1.00; validation study = 0.94) show appropriate control for inflation due to population structure. B) Regional plot of discovery study associations indicating both the significance level and the recombination rate at the OR5A1 LD-band. The previously discovered D183N variant (rs6591536; OR5A1) also changes perceived intensity of β-ionone in our populations: C) discovery cohort, D) validation cohort (high concentration). The x-axis is ordered left-to-right for increasing number of variant alleles, with population frequency indicated below the genotype.
Table 1.
Variant age and effect of derived allele on odor perception.
In a literature review, derived alleles corresponded with decreased sensitivity to odor in 21 out of 29 cases, and 11 out of 13 cases with functional validation. All but two variants predate the estimated ages of the East Asian and European population divergences.
Table 2.
Concentrations of Odors from the Discovery and Validation Studies.