Does a merging event always precede AGN activity? In this starting PhD project, we will try to address this question by quantifying the distortions and interaction signatures that can be detected on a sample of AGN galaxies from the ACS-COSMOS survey. In order to do this, we will investigate and combine all the different available galaxy descriptors and develop an automatic morphological classification method. The goal is to find which combination of parametric and non-parametric estimators yield a quantitative indicator of interaction on the host galaxy, and also define the strength level of these signatures that we are able to detect.