Nica et al. [5] proposed a methodology to rank the SNPs with an influence on two traits based on the residual association conditional on the most associated SNP. By comparing the GWAS SNP score with all other SNPs in the associated region, this method accounts for the local LD structure. However, this is not a formal test of a null hypothesis for, or against, colocalisation at the locus of interest. A formal test of colocalisation has been developed in a regression framework. This is based on testing a null hypothesis of proportionality of regression coefficients for two traits across any set of SNPs, an assumption which should hold whenever they share causal variant(s) [12], [13]. No assumption is made about the number of causal variants, although the method does assume that in the case of multiple causal variants, all are shared. Both the ranking method and proportionality testing share the drawback of having to specify a subset of SNPs to base the test on, and Wallace [14] shows that this step can generate significant biases. The main sources of bias
