Stanford University, Fujitsu Laboratories Ltd.

Molecular dynamics methods utilizing a Generalized Bennett acceptance ratio method (BAR; M. Shirts et al. 2003) present themselves as an ideal alternative for modeling the true energetics at work in ligand binding events, including the complex effects of solvation and desolvation. Unfortunately, these methods can be extremely computationally intensive due to the large number of trajectories, which must be solved in order to model the system.

Scientists at Fujitsu Labs, working with collaborators at Stanford University, addressed this problem utilizing the BioServer. In order to get accurate binding energy, they used the Amber 1999 force field for the FKBP receptor and a new general Amber force field (GAFF; J. Wang et al. 2004) for a set of nine ligands for which experimental inhibition data were available. They found that the parallel configuration of the BioServer enabled calculations in days that would take a standard cluster configuration months-to-years to complete. Moreover, when the modeled results were compared with experimental inhibition constants, a very tight agreement was seen (± 1 kcal ).

Hideaki Fujitani¹, Yoshiaki Tanida¹, Masakatsu Ito¹, Michael R. Shirts², Christopher D. Snow², Guha Jayachandran² and Vijay S. Pande<sup>2

1</sup> Fujitsu Laboratories Ltd., 10-1 Morinosato- Wakamiya, Atsugi, Kanagawa, Japan
Tel: +81-46-250-8234, Fax: +81-46-250-8844, e-mail: fjtani@labs.fujitsu.com
² Department of Chemistry, Stanford University, Stanford, CA 94305-5447, USA