Spatial clustering of molecular dynamics trajectories in protein unfolding simulations

Molecular dynamics simulations is a valuable tool to study protein unfolding in silico. Analyzing the relative spatial position of the residues during the simulation may indicate which residues are essential in determining the protein structure. We present a method, inspired by a popular data mining technique called Frequent Itemset Mining, that clusters sets of amino acid residues with a synchronized trajectory during the unfolding process. The proposed approach has several advantages over traditional hierarchical clustering.The authors acknowledge the support of the "Fundacao para a Ciencia e Tecnologia,Portugal, and the program FEDER, through grant PTDC/BIA-PRO/72838/2006 (to PJA and RMMB) and the Fellowships SFRH/BPD/42003/2007(to PGF) and SFRH/BD/16888/2004 (to CGS). We thank the Center for Computational Physics, Departamento de Fisica, Universidade de Coimbra, for the computer resources provided for the MD simulations