Tests of aggregate rare variant association applied to a multiethnic sequencing study. A. D. Ablorh1, S. Lindstrom1, C. A. Haiman2, B. E. Henderson2, L. Le Marchand3, S. Lee4, D. O. Stram2, P. Kraft1,4 1) Epidemiology, Harvard School of Public Health, Boston, MA; 2) Department of Preventive Medicine, Keck School of Medicine and Norris Comprehensive Cancer Center, University of Southern California, Los Angeles, California; 3) Epidemiology Program, University of Hawaii Cancer Research Center, Honolulu, HI; 4) Biostatistics, Harvard School of Public Health, Boston, MA.
For many complex diseases, the genetic basis of susceptibility has not been fully explained and it is possible that rare variants with a minor allele frequency (MAF)0.005, contribute to the remaining genetic heritability. Several association methods have been proposed for aggregating information from individual rare variants in order to maximize power. However, these have been proposed in the context of studies conducted in a single ethnicity. It remains an open question how to best combine rare variants in association testing across ethnicities in a multi-ethnic study. Since causal variants unique to one population may be harder to detect but causal variants that are common to all may reveal themselves, performance of rare variant association methods may differ in multi-ethnic compared to single-ethnic study populations. We selected four statistical approaches for comparison. Three involve first applying the Combined Multivariate Collapsing Method (CMC) within each population separately. The first tests whether the common collapsed effect is non-zero using fixed effects meta-analysis; the second tests whether any of the unconstrained collapsed effects is non-zero; the third, a modification of MANTRA, constrains the collapsed effects according to population genetic similarity. The last approach is an extension of the Sequence Kernel Algorithm Test (SKAT) that combines results across populations, MetaSKAT. We applied all four methods to a targeted sequencing study where we sequenced twelve breast cancer susceptibility loci in 1560 women (771 cases and 789 controls) from the Nurses Health Study (NHS) and 2107 women (1542 cases and 1565 controls) from the Multiethnic Cohort (MEC). Sequencing allows us to characterize variants across a spectrum of allele frequencies much lower than those previously explored. Our study population includes women from four ethnic groups (33% European-, 20% African-, 19% Japanese-, and 27% Latin-American). 87.3% of sequenced variants were rare (MAF0.005), and 70.2% were population private. We present the results of meta-analysis of rare genetic variation using the four statistical approaches in a multi-ethnic population, and discuss the implications for design and analysis of future studies.
You may contact the first author (during and after the meeting) at