Fig. 2.

Overview of results for investigated protein families. To understand the effect of residue coevolution in homo-oligomerization, we investigate a total of 1,984 protein families (PF) from the Pfam database. They fulfill the requirements of sufficient sequence information and a known monomeric structure, which is a requirement to differentiate coevolving intraprotein contacts (i.e., tertiary contacts) from putative interprotein contacts (i.e., quaternary contacts). Larger assemblies have stronger signals, because only some residues forming the interface strongly coevolve. We thus subdivide the PF into two classes pending their known interface sizes. A total of 750 PF have a large known interface in their biological assembly. For 372 PF of these total PF, we can identify this interface through strongly coevolving residue pairs based purely on sequences. For the other 1,234 PF, no large biomolecular assemblies are known. A total of 1,088 PF, however, form a known small interface, and this interface can be identified for 189 PF. For the 146 PF possessing no known interface, the large majority do not show strongly coevolving residue pairs. For 24, however, we find strongly coevolving pairs indicating a possible biological homo-oligomeric assembly.