Fig. 3.

Markov diffusion clustering of QAA shows ubiquitin motions involved in binding substrates: The 30 dimensional space determined form QAA is used to construct a set of meta-stable states that are energetically accessible. From a group of 10 000 conformers, we show how the network is modeled with the adjacency matrix C₀ shown here. The Markov diffusion produces a total of 78 macro-states at level 4 of the hierarchy. To illustrate the extremum points in the network, we depict two representative clusters (A and B shown in red and green respectively) representing changes within the binding regions of ubiquitin. The primary binding region is indicated by R1.