Computational methods for the prediction of bacterial pathogen-host protein protein interactions.

For an understanding of bacterial virulence it is necessary to identify the molecular key players in microbe-host interactions. In this work, a taxonomically universal approach to predict secreted bacterial effector proteins is presented. To make predicted effectomes accessible, we devised a comprehensive web resource. On the host sided part, the potential of domain based methods to predict pathogen-host protein interactomes was explored. Furthermore, it was analyzed to which extent the repertoire of effector candidates in the genome of a bacterial organism enables prediction of a host-interacting phenotype