Supporting Information

Files in this Data Supplement:

Supporting Table 3
Supporting Figure 2
Supporting Figure 3
Supporting Figure 4
Supporting Table 4
Supporting Text

Fig. 2. Comparison of SASA-only (A), SAV-only (B), and full SPT (C) model forces with total explicit solvent force component using the optimal parameters in Table 2.

Fig. 3. Sensitivity of the mean squared error (c²) to solvent probe radius (s_s) for the WCA (solid line) and 6/12 (dashed line) versions of the full model (Eq. 18).

Fig. 4. Comparison of explicit and implicit solvent forces using the optimal parameters in Table 1 for the WCA (A) and 6/12 (B) models.

Supporting Text

Timing Information for Nonpolar Calculations

The IFABP implicit solvent nonpolar calculations were performed by using GCC-compiled code (default optimization) on a single 2.8-GHz Intel Pentium 4 processor with 2 GB RAM. All times are approximate results obtained from multiple runs on the IFABP system. The average execution time of the integral code was 7 s. Each single-point SAV evaluation required 0.6 s; each SAV derivative took 1 s. The single-point SASA calculations were 1 s in duration, whereas SASA derivatives required 6 s. Note that all of these run times are "wall clock" and include time for reading structural data and parameters and writing output; these steps could be avoided in higher-throughput applications such as dynamics simulations. The integral code has not currently been optimized for speed. Finally, much more advanced methods for volume calculations (and derivatives) are available (1, 2); use of techniques such as these could further improve the efficiency of the SASA and SAV computations.

1. Edelsbrunner, H. & Koehl, P. (2003) Proc. Natl. Acad. Sci. USA 100, 2203-2208.

2. You, T. & Bashford, D. (2005) J. Comput. Chem. 16, 743-757.