Web-based, participant-driven studies yield novel genetic associations for common traits

Abstract

Despite the recent rapid growth in genome-wide data, much of human variation remains entirely unexplained. A significant challenge in the pursuit of the genetic basis for variation in common human traits is the efficient, coordinated collection of genotype and phenotype data. We have developed a novel research framework that facilitates the parallel study of a wide assortment of traits within a single cohort. The approach takes advantage of the interactivity of the Web both to gather data and to present genetic information to research participants, while taking care to correct for the population structure inherent to this study design. Here we report initial results from a participant-driven study of 22 traits. Replications of associations (in the genes OCA2, HERC2, SLC45A2, SLC24A4, IRF4, TYR, TYRP1, ASIP, and MC1R) for hair color, eye color, and freckling validate the Web-based, self-reporting paradigm. The identification of novel associations for hair morphology (rs17646946, near TCHH; rs7349332, near WNT10A; and rs1556547, near OFCC1), freckling (rs2153271, in BNC2), the ability to smell the methanethiol produced after eating asparagus (rs4481887, near OR2M7), and photic sneeze reflex (rs10427255, near ZEB2, and rs11856995, near NR2F2) illustrates the power of the approach.

Figure 1. Web-based accrual of genotype and phenotype data via a personal genetic information service.

(A) Individual participant signs up for service via web; (B) Participant receives sample collection kit; (C) Participant consents, via web, to use of anonymized genotype and survey responses for research; (D) Participant sends saliva sample to contracted lab; (E) Participant logs on to service web site with option to respond to one or more surveys, prior to having access to personal genetic data; (F) Laboratory processes sample, generating data for single nucleotide polymorphisms (SNPs); (G) Encrypted genotype data transferred from laboratory to secure server; (H) Participant logs on to service web site with option to access personal genetic data (both raw genotypes and customized reports); (I) Participant logs on and has the option to respond to one or more surveys; (J) New genetic reports posted, new surveys posted; (K) Genotype data and survey responses for individuals, coded and stripped of individually identifying information, are transferred to research team. Shaded boxes indicate participant actions; clear boxes indicate lab or service actions. Dashed boxes indicate optional participant actions within framework of service access.

Figure 2. Manhattan plots for new associations.

Shown are (A) hair curl, (B) asparagus anosmia, (C) photic sneeze reflex, and (D) freckling. Plots show scores ( p-values) for all SNPs by physical position. All plots are trimmed at a maximum score of 15. For regions with a more significant association, the strongest score in that region is shown above the region.

Figure 3. Manhattan plots for replications.

Shown are (A) hair color, (B) red hair, (C) blue to brown eye color, and (D) green versus blue eye color. Plots show scores ( p-values) for all SNPs by physical position. All plots are trimmed at a maximum score of 15. For regions with a more significant association, the strongest score in that region is shown above the region.

Figure 4. Quantile-quantile plots for new associations and replications.

Shown are (A) hair curl, (B) asparagus anosmia, (C) photic sneeze reflex, (D) freckling, (E) hair color, (F) red hair, (G) blue to brown eye color, and (H) green versus blue eye color. All plots are trimmed at a maximum score of 15 to better show details. Approximate 95% CIs are indicated in blue. The red line passes through the median p-value.

Figure 5. Bayes factors for genotyped and imputed SNPs for hair curl around TCHH.

Plotted are Bayes factors for all SNPs in HapMap in the region. Red/blue circles represent typed SNPs, magenta/green squares imputed SNPs; Red/magenta points have log Bayes factors over 6. Genes are marked in the top track with gene names inside the figure. Linkage disequilibrium (relative to the SNP with the largest Bayes factor) is plotted in the middle track.

Figure 6. Bayes factors for genotyped and imputed SNPs for hair curl around WNT10A.

For details, see Figure 5.

Figure 7. Bayes factors for genotyped and imputed SNPs for hair curl around OFCC1.

For details, see Figure 5.

Figure 8. Bayes factors for genotyped and imputed SNPs for asparagus anosmia around OR2M7.

For details, see Figure 5.

Figure 9. Bayes factors for genotyped and imputed SNPs for two photic sneeze associations.

Regions shown are (A) near ZFHX1B/ZEB2 on 2q22.3 and (B) near NR2F2 on 15q26.2. For details, see Figure 5.

Figure 10. Bayes factors for genotyped and imputed SNPs for freckling around BNC2.

For details, see Figure 5.

Figure 11. Association over time for two SNPs and two traits.

(A) Association between rs4481887 and asparagus anosmia, steadily increasing in certainty. (B) Association between rs6650382 and photic sneeze: a very promising initial result, but , the log odds ratio, regressed towards 0. Both traits were assessed in the same survey, so they had approximately the same number of responses at all points in time. Scores ( p-values) and regression coefficients are plotted using the genotype and phenotype data that was available at various points in time. The red line indicates our significance threshold of 8.4.