Schematic illustrations of a defective clique and how the concept evolved. (A) A defective clique in a protein interaction network. KP and KQ are both (k + 1)-cliques, with k overlapping vertices (i.e. clique K). The dashed edge between proteins P and Q corresponds to a predicted interaction. KPQ is a defective clique with a missing edge PQ. (B) The decomposition of the defective clique (KPQ) into the union of two overlapping cliques (KP and KQ). (C) Generalized defective cliques. In general, a defective clique consists of two cliques: K {cup} KP and K {cup} KQ. There are two parameters to determine a defective clique: k, the size of the overlapping subclique (i.e. K); l, the size of the non-overlapping subcliques (i.e. KP {cup} KQ). In the defective clique K {cup} KP {cup} KQ, the dashed edges between subcliques KP and KQ correspond to predicted interactions. [from H. Yu, A. Paccanaro, V. Trifonov, M. Gerstein (2006). Predicting interactions in protein networks by completing defective cliques. Bioinformatics 2006 Apr 1;22(7):823-9]