For the vast majority of naturally occurring, small, single-domain proteins, folding is often described as a two-state process that lacks detectable intermediates. This observation has often been rationalized on the basis of a nucleation mechanism for protein folding whose basic premise is the idea that, after completion of a specific set of contacts forming the so-called folding nucleus, the native state is achieved promptly. Here we propose a methodology to identify folding nuclei in small lattice polymers and apply it to the study of protein molecules with a chain length of N = 48. To investigate the extent to which protein topology is a robust determinant of the nucleation mechanism, we compare the nucleation scenario of a native-centric model with that of a sequence-specific model sharing the same native fold. To evaluate the impact of the sequence's finer details in the nucleation mechanism, we consider the folding of two non-homologous sequences. We conclude that, in a sequence-specific model, the folding nucleus is, to some extent, formed by the most stable contacts in the protein and that the less stable linkages in the folding nucleus are solely determined by the fold's topology. We have also found that, independently of the protein sequence, the folding nucleus performs the same 'topological' function. This unifying feature of the nucleation mechanism results from the residues forming the folding nucleus being distributed along the protein chain in a similar and well-defined manner that is determined by the fold's topological features.