GWAS of wing shape in Drosophila melanogaster

     The Drosophila wing has come to be a great model system for genetic and evolutionary analysis of shape for a number of reasons. Wings are essentially 2D, WYSIWYG, fast to measure structures; wing shape is evolutionarily conserved and yet its variation is relatively high-dimensional; and the fact that they're attached to a model organism means that we know a great deal about their genetics and development, including spatially explicit information like gene expression domains at various developmental stages. All of these features make wing shape an excellent subject for a high-dimensional genome-wide association study (GWAS), which we will be soon completing.
     Unlike single-variable GWAS, the core emphasis of a multivariate approach is not on inference---determining whether a SNP is a significant QTN--but on estimation of effects. These multivariate estimates take the form of vectors that describe the contributions of each original measurement to the difference between two genotypes. Therefore, SNP vectors are hypotheses about the precise pleiotropic effects of a nucleotide substitution. These hypotheses can then be validated in an independent population not only by asking whether the same SNP produces the same effect---the only validation option for univariate data--but also by quantifying the degree of correlation between sample estimates.
     Similarly, characterizing SNP effects as vectors facilitates performing comparisons among SNP effects, which we are using to generate and test functional interaction hypotheses, many of which are likely to be novel. On the whole, our work is showing how pursuing a GWAS on multiple dimensions opens a universe of options for comparison that enormously enhances our ability to visualize, interpret, explain, and validate putative QTNs.