Figure 2

Statistical interactions between genetic factors with respect to multiple depression-related phenotypes in persons with high exposure to recent negative life events. A Bayesian interaction score was computed for all possible variable pairs in order to measure their joint occurrence in relevant models with respect to multiple depression-related phenotypes. The individual relevance of each SNP is shown in the outer grey ring with red columns having a height proportional to their relevance. Possible interactions between SNPs are represented by curved red lines between corresponding columns with a width proportional to the likeliness of the interaction. Interactions related to SNPs with low individual relevance (e.g. CB1 and HTR2A SNPs, denoted with dashed orange lines) can be neglected, as the interaction score is sufficient only in case of variables with moderate or high relevance. Results indicated two interactions: a strong interaction between rs6265 (BDNF) - rs6295 (HTR1A) and a mild interaction between 5-HTTLPR (SLC6A4) – rs8836 (GALR2). The rs6265 (BDNF) - rs6295 (HTR1A) interaction was also observed in the GLM analysis of models (see Table 2). In contrast, the interaction of 5-HTTLPR (SLC6A4) – rs8836 (GALR2) was not revealed by the GLM analysis, which indicates that this interaction is of a different type, and only detectable by systems-based methods. Note that results shown in Fig. 1 suggest that the 5-HTTLPR (SLC6A4) and rs8836 (GALR2) SNPs have separate main effects with respect to depression.