Using Data Mining Techniques to Probe the Role of Hydrophobic Residues in Protein Folding and Unfolding Simulations

The protein folding problem, i.e. the identification of the rules that determine the acquisition of the native, functional, three-dimensional structure of a protein from its linear sequence of amino-acids, still is a major challenge in structural molecular biology. Moreover, the identification of a series of neurodegenerative diseases as protein unfolding/misfolding disorders highlights the importance of a detailed characterisation of the molecular events driving the unfolding and misfolding processes in proteins. One way of exploring these processes is through the use of molecular dynamics simulations. The analysis and comparison of the enormous amount of data generated by multiple protein folding or unfolding simulations is not a trivial task, presenting many interesting challenges to the data mining community. Considering the central role of the hydrophobic effect in protein folding, we show here the application of two data mining methods – hierarchical clustering and association rules – for the analysis and comparison of the solvent accessible surface area (SASA) variation profiles of each one of the 127 amino-acid residues in the amyloidogenic protein Transthyretin, across multiple molecular dynamics protein unfolding simulations.