Abstract for presentation at 11th International Congress of Human Genetics

Power of TDT and AFBAC in genome-wide association studies

  • Xueying Liang, Center for Human Genetics Research and Department of Molecular Physiology and Biophysics, Vanderbilt University, United States
  • Dr Jacob McCauley, Center for Human Genetics Research and Department of Molecular Physiology and Biophysics, Vanderbilt University, United States
  • Mr Justin Giles, Center for Human Genetics Research and Department of Molecular Physiology and Biophysics, Vanderbilt University, United States
  • Dr Jonathan Haines, Center for Human Genetics Research and Department of Molecular Physiology and Biophysics, Vanderbilt University, United States
  • IMSGC, United States

    • Genome-wide association studies are underway as a promising approach for identifying the genes that underlie common diseases. Family-based association tests combined with high density SNP arrays have advantages for mapping disease genes. The AFBAC (affected family-based controls) method provides an unbiased estimate of the overall population (control) marker alleles using non-transmitted parental allele and includes homozygous parents in the analysis. In contrast, homozygous parents are not informative in TDT method. We performed simulations to test the power of TDT and AFBAC on parent-child trio data in genome-wide association studies and to see if either method had advantages under a variety of genetic models relevant to multiple sclerosis.
    We simulated 500,000 SNPs with 1000 trios. We simulated two disease loci such that the joint effect of two disease loci was controlled at various Odds Ratios (OR). Three combinations of disease allele frequencies (0.1/0.4;0.2/0.3;0.6/0.6) were used to simulate rare/common disease alleles. Dominant, recessive and additive inheritance models were applied to all combinations of disease allele frequencies. Each of these 9 models had 1000 replicates. TDT and AFBAC were used to analyze all datasets.
    Our data demonstrate that TDT and AFBAC have similar power regardless of the significance level used. Both methods have almost 100% power to detect either one of two disease loci in most of the models when the OR=2.0. False positive rates at α=0.05 were 5.2% in both methods. In dominant and additive models, the allele frequency doesn’t affect power significantly. However, it is very hard to detect an effect for a recessive model with a low allele frequency. We suggest that TDT can be used to detect association in a certain population at reduced computation time and effort and provide competitive power to the AFBAC. More details of the study, such as the effect of genotyping error and missing genotypes, are under investigation.

    Conference Organiser - ICMS Pty Ltd