We have investigated the self-assembly scenario of patchy colloidal particles in a two-dimensional system. The energetically most favourable ordered particle arrangements have been identified via an optimization tool that is based on genetic algorithms. Assuming different simple models for patchy colloidal particles, which include binary mixtures as well as attraction and repulsion between the patches, we could identify a broad variety of highly non-trivial ordered structures. The strategies of the systems to self-assemble become evident from a systematic variation of the pressure: (i) saturation of patch bonds at low pressure and close packing at high pressure and (ii) for intermediate pressure values, the strategy is governed by a trade-off between these two energetic aspects. The present study is yet another demonstration of the efficiency and the high reliability of genetic algorithms as versatile optimization tools.