The formation of Pt-Ru nanoclusters (NCs) by sequential deposition of Pt and Ru on a periodically rumpled graphene sheet supported on Ru(0001) is analyzed by atomistic-level modeling and kinetic Monte Carlo simulations. The "coarse-scale" periodic variation of the adsorption energy of metal adatoms across the graphene sheet directs the assembly of NCs to a periodic array of thermodynamically preferred locations. The modeling describes not only just the NC densities and size distributions, but also the composition distribution for mixed NCs. A strong dependence of these quantities on the deposition order is primarily related to different effective mobilities of Pt and Ru on the supported graphene.