Faster coevolution detection of proteins using maximum similar cliques.

We present a new and faster algorithm for predicting coevolving proteins from pairs of protein distance matrices that are free to differ in size and scale, based on the initial work in [3]. The new algorithm has a more effective pruning strategy for possible groupings of proteins, and a more precise test for co-evolution. Its theoretical worst-case time complexity is O(N4 · 32/3N), and in practice, exhibits execution times that are on the order of 100 times faster than the original.