Multiple Imputation in Genetic Linkage Analysis

• Activity Number: 318
• Type: Contributed
• Date/Time: Wednesday, August 14, 2002 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section*
• Author(s): Darlene Goldstein*+ and Sandrine Dudoit
• Affiliation(s): University of California, Los Angeles and University of California, Berkeley
• Keywords: statistical genetics ; multiple imputation ; linkage analysis ; score test ; allele sharing

Abstract

Dudoit and Speed (Biostatistics 1:1-26) proposed a likelihood-based test in the recombination fraction (theta) for genetic linkage of a trait to a marker. Simulation studies showed that this test is more powerful than commonly employed nonparametric tests, while also being robust to a large range of model misspecification (Goldstein, Dudoit, Speed; Genetic Epidemiology 19:S85-S91, 20:415-431). The test uses identity by descent (IBD) information on pairs of relatives; that is, the number of shared alleles inherited from a common ancestor must be known. Often, however, the available genotype data is limited to information on the allele states (identity by state, or IBS) and, thus, there is some information missing.

We extend the applicability of this test by using multiple imputation to estimate the linkage score statistic with IBS (i.e., missing IBD) data. The performance of the method is assessed for simulated genotype data on relative pairs (siblings, half-sibs, cousins, avuncular, and grandparent-grandchild).