Case-control association testing in the presence of unknown relationships

Abstract

Genome-wide association studies result in inflated false-positive results when unrecognized cryptic relatedness exists. A number of methods have been proposed for testing association between markers and disease with a correction for known pedigree-based relationships. However, in most case-control studies, relationships are generally unknown, yet the design is predicated on the assumption of at least ancestral relatedness among cases. Here, we focus on adjusting cryptic relatedness when the genealogy of the sample is unknown, particularly in the context of samples from isolated populations where cryptic relatedness may be problematic. We estimate cryptic relatedness using maximum-likelihood methods and use a corrected chi(2) test with estimated kinship coefficients for testing in the context of unknown cryptic relatedness. Estimated kinship coefficients characterize precisely the relatedness between truly related people, but are biased for unrelated pairs. The proposed test substantially reduces spurious positive results, producing a uniform null distribution of P-values. Especially with missing pedigree information, estimated kinship coefficients can still be used to correct non-independence among individuals. The corrected test was applied to real data sets from genetic isolates and created a distribution of P-value that was close to uniform. Thus, the proposed test corrects the non-uniform distribution of P-values obtained with the uncorrected test and illustrates the advantage of the approach on real data.

Figure 1

Pedigree for simulation study: black is affected and white is unaffected.

Figure 2

Box plots of estimated kinship coefficients. A: the simulated sample of Scenario I using 50 and 400 microsatellite markers. B: CEPH families using 500, 5,000 and 16,977 SNP markers.

Figure 3

Quantile-quantile plots of χ² test and corrected χ² tests for 50 markers (dashed line) and 400 markers (solid line) in the simulated sample of Scenario I. A: Classical χ² test. B: Corrected χ² test with actual kinships. C: Corrected χ² test using estimated kinships. D: Corrected χ² test using pedigree-based kinships.

Figure 4

Quantile-quantile plots of χ² test and corrected χ² tests in the simulated sample of Scenario II.

Figure 5

Comparison of kinship coefficients: actual kinships vs. estimated, pedigree-based and posterior kinships in the simulated sample of Scenario III.

Figure 6

Estimated k-coefficients of pairs in (A) Guam and (B) Kosrae samples. Cumulative distribution of estimated kinship coefficients of cases and controls in the (C) Guam and (D) Kosrae data sets. Quantile-quantile plot of p-values of χ² test and corrected χ² test in (E) Guam and (F) Kosrae samples. In the Kosrae study, ϕ is the pedigree-based kinship coefficient and ϕ̂ is the estimated kinship coefficient.