The interpretation of the posterior probabilities requires caution. For example, a low PP4 may not indicate evidence against colocalisation in situations where PP3 is also low. It may simply be the result of limited power, which is evidenced by high values of PP0, PP1 and/or PP2. Moreover, a high PP4 is a measure of correlation, not causality. To illustrate this, one can consider the relatively common situation where a single variant appears to affect the expression of several genes in a chromosome region (as observed, for example, in the region surrounding the SORT1 gene). Several eQTLs will be colocalised, both between them and with the biomarker of interest. In this situation one would typically expect that a single gene is causally involved in the biomarker pathway but the colocalisation test with the biomarker will generate high PP4 values for all genes in the interval.
