On the Meaning of Averages in Genome-wide Association Studies: What Should Come Next?

Abstract

Identifying the association between phenotypes and genotypes is the fundamental basis of genetic analyses.
Although genomic technologies used to generate data have rapidly advanced within the last 20 years, the statistical models used in genome-wide associations studies (GWAS) to analyze these data are still predominantly based on the model developed by Fisher more than 100 years ago. The question is, does Fisher’s theory need to be replaced or improved, and if so, what should come next? The theory developed by Fisher was inspired by the
field of probability. To make use of probability not only did Fisher have to assume valid a number of questionable
hypotheses, but he also had to conceptually frame genotype-phenotype associations in a specific way giving
primordial importance to the notion of average. However, the “average” in probability results from the notions
of “imprecision” or “ignorance”. After reviewing the historical emergence and societal impact of probability as
a method, it is clear what is needed now is a new method acknowledging precision in measurements. That is, a
method that does not rely on categorizing or binning data.