The original version of our PPM program was modified to optimize positions of molecules in the lipid bilayer with the new solvation model. The computation requires only energy functions and parameters described in equations (142) and Tables S1, S3, S4. A solute molecule was considered as a rigid body whose spatial position was defined by three independent variables: two rotation angles and one translation along the bilayer normal (φ, τ and d, respectively). Transfer energy (equation 19) was optimized by combining grid scan and Davidon-Fletcher-Powell method for local energy minimization. Derivatives of transfer energy with respect to rigid body variables of the molecule were analytically calculated, as described previously22 (link). Derivatives of energy with respect to z were calculated as finite differences with step of 0.01 Å. The hydrophobic thickness of TM proteins was optimized by grid scan for location of the hydrocarbon boundary (ZHDC in eq. 2) with step of 0.05 Å. All other peaks of the lipid and water distributions were shifted accordingly during the optimization to remain at the same distance from the hydrocarbon boundary as in DOPC.
The program uses as input only a set of coordinate files in the PDB format. Unlike the previous version, it allows an automatic determination of transmembrane secondary structures without using any external software. The dipole moments and standard pKa values of different groups are included in the library of amino acid residues or directly in the PDB files for small molecules.