From Buccal to Bonferroni in Genetically Informative Prevention Research III: Operationalizing and Analyzing Genetic Variance from SNP to GWAS and in-Between

This final presentation in the Buccal to Bonferroni series will present several methods of operationalizing genetic variance and analyzing genetic by environment research questions.

A primary question researchers must consider is how to represent genetic variance in their gene x intervention study. This choice is not straightforward. Genetic variance can be characterized on the level of single nucleotide polymorphisms (SNPs), Variable Number of Tandem Repeats (VNTRs), haplotypes, summative gene-scores, and assessments across the genome (genome-wide association studies, GWAS) and epi-genome (EWAS). Each of these methods will be defined and examples given. In addition, strengths and weaknesses of these approaches and their applicability to projects that differ substantially on sample characteristics will be presented.

For example, because they can evaluate associations between literally millions of SNPs and phenotypes, GWAS approaches require substantial statistical power and are therefore best suited for large samples that can adequately adjust for risks associated with type 1 error. Moreover, given the many testable combinations of phenotypes and genotypes that can be generated by many measurement-rich intervention studies, G x E analyses of these data sets can be plagued by multiple testing. Nonetheless, given what molecular genetic research has learned about how these specific candidate genes influence phenotypes that prevention researchers care about, ranging from substance use to depression, single-gene x intervention approaches can provide insight regarding the socio-biological pathways through which interventions affect behavioral outcomes. A mid-level approach to gene x intervention research is offered by the use of multi-locus gene scores. As they can account for more variance than single-genes, the use of these scores is becoming increasingly popular. The poster will explain the two types of gene scores that are commonly used, neurotransmitter specific (e.g., dopamine gene scores) and phenotype-specific (e.g., multi-locus genetic smoking risk scores), how to develop these different types of scores, and present an overview of their strength and weaknesses.

In addition to the above, we will present an overview of population structure, how it can lead to spurious findings, and different approaches to addressing this confound.

Careful attention to the details found in these three posters will prevent future prevention researchers from having to “reinvent the wheel” when designing, executing, and analyzing genetically informed studies.