Biostatistics 666: Association Tests in Structured Populations

Overview

This lecture introduces the problem of population stratification in the context of genetic association studies. It also discusses introduces several possible strategies for detecting and handling population structure, specifically:

• Q-Q plots as a diagnostic tool in genetic association studies.
• Using family based association tests, such as the TDT, to avoid spurious association even when a single marker is tested.
• Using the genomic control method to controling the overall inflation of test statistics, when many markers are tested.
• Using genotype based matching approaches to model population structure more explicitly.

Lecture Notes

Slides in PDF Format

Recommended Reading

• Genomic Control for Association Studies
  • Devlin and Roeder (1999) Biometrics 55:997-1004
  • Pritchard and Rosenberg (1999) Am J Hum Genet 65:220-228

• Methods for Inferring Population Structure
  • Pritchard, Stephens and Donnelly, 2000. Genetics 155: 945-959

• Principal Components Analysis
  • Price et al (2006) Nature Genetics 38:904-909

• Transmission Disequilibrium Tests
  • Spielman et al (1993) Am J Hum Genet 52:506-16 (trios)
  • Curtis (1997) Ann Hum Genet 61:319-33 (sibling pairs)