An explanation of interaction data types and their representation. Experiments that can be used to detect interactions may produce binary data (panels A-C) or n-ary data (panels D-F). N-ary data is commonly represented using one of three different models: the spoke model (panels G-H), the matrix model (panels I-J) or the bipartite model (panels K-L). Some n-ary data is represented using a spoke model and an attempt can be made to detect these records in MITAB files and reconstruct the list of component proteins in a potential complex (panel M). These lists are referred to as regenerated complexes in this paper. The full explanation of this Figure is provided in the Methods section.