Researching Genetic Versus Nongenetic Determinants of Disease: A Comparison and Proposed Unification

Abstract

Research standards deviate in genetic versus nongenetic epidemiology. Besides some immutable differences, such as the correlation pattern between variables, these divergent research standards can converge considerably. Current research designs that dissociate genetic and nongenetic measurements are reaching their limits. Studies are needed that massively measure genotypes, nongenetic exposures, and outcomes concurrently.

The False-positive to False-negative Ratio in Epidemiologic Studies

“…Very large cohort studies and cohort consortia are available and several hundreds or even a few thousand exposures can be measured and analyzed cumulatively in a standardized, consistent manner. 57,58 However, to date this is not happening in mainstream epidemiology. Table 2 Even if one were to work in fields with R ϭ 1 (1 genuine effect tested for every null effect tested)-ie, conducting research on topics that are already very well studied and on risk factors that already have very strong prior evidence-the FP:FN ratio would still be 3.2:1 unless bias is accounted for.…”

Strengthening the reporting of genetic risk prediction studies (GRIPS): explanation and elaboration

“…Adjusted estimates should be presented next to the unadjusted estimates, so that readers are able to judge the extent to which the findings change by the inclusion of other risk factors in the model. This is particularly relevant for models that combine genetic and nongenetic risk factors, because nongenetic risk factors can be intermediate factors in the biological pathway [39] and many nongenetic risk factors have complex correlation patterns [67,68]. Note that several studies have presented adjusted effect sizes for genetic variants (e.g.…”

Expectations, validity, and reality in omics

“…Even if some of the new measurements reach astonishing accuracy (e.g., genotyping error in genome-wide association studies can be reduced to !0.1%), most research questions also require measuring other variables (e.g., environmental exposures or clinical outcomes), where the measurement error can be 100-fold higher [24]. Moreover, the whole genre of biases that affect case-control, cohort, or other epidemiological designs remains an issue, even if the measurements are made on the newest omics platform.…”