Hypothesis-driven candidate gene association studies: practical design and analytical considerations

Abstract

Candidate gene association studies (CGAS) are a useful epidemiologic approach to drawing inferences about relations between genes and disease, especially when experimental data support the involvement of specific biochemical pathways. The value of CGAS is apparent when allele frequencies are low, effect sizes are small, or the study population is limited or unique. CGAS is also valuable for validating previous reports of genetic associations with disease in different populations. Despite the many advantages, the information generated from CGAS is sometimes compromised because of either inefficient study design or suboptimal analytical approaches. Here the authors discuss issues related to the study design and statistical analyses of CGAS that can help to optimize their usefulness and information content. These issues include judicious hypothesis-driven selection of biochemical pathways, genes, and single nucleotide polymorphisms, as well as appropriate quality control and analytical procedures for measuring main effects and for evaluating environmental exposure modifications and interactions. A study design algorithm using the example of DNA repair genes and cancer is presented for purposes of illustration.

Figure 1.

Flow diagram of the gene/single nucleotide polymorphism (SNP) selection process used in a candidate gene association study. Seven sequential steps identify candidate genes, prioritize them on the basis of hypothesis, select the most parsimonious combination of functional and tagging SNPs, and pare the final SNP count to the constraints of the platform. In this case, the platform is the Illumina GoldenGate chip (Illumina, Inc., San Diego, California), which has 1,536 SNP genotyping slots. See the “Application” section of the text for a full explanation of the gene/pathway prioritization algorithm. MAF, minor allele frequency; NMSC, nonmelanoma skin cancer. (The Tagger SNP selection program was produced by the Broad Institute, Cambridge, Massachusetts.)