Step 2: Preparation and Submission of Input Files for the QM Procedure
For MFCC, after generating fragments and conjugating caps, all fragments (–ligand), caps (–ligand), and ligand molecules are evaluated by QM calculations (if N fragments have been generated for a protein, the number of QM input files will be 4N – 1). Because the first (last) fragment and conjugated cap(s) share the same structure, they are not calculated, and the number of input files is 4N − 9 (large, even for a moderately sized protein). Therefore, automated preparation of QM input files is desirable. In the implementation of GridMol version 2.0, users can select one QM package (e.g., Gaussian) and provide parameters to allow automatic generation of the input files.
For FMO, because the process of generating a GAMESS input file is not straightforward, users must create a complex input file while consulting a manual. In the implementation described here, we analyze each fragment’s characteristics and perform an automated generation of the input file. Bonds are fractioned electrostatically, and bond-specific electron pairs remain intact. Each atom in a molecule is grouped into a unique fragment, with the charge of each fragment assigned according to its electrons, which are assigned heterolytically during molecule fragmentation. In FMO, the sp2 or sp3 hybrid orbitals are used for bond detachment. In this study, we created orbitals for sp3 C, sp3 N, and sp2 C for frequently used Gaussian-based functions [STO-3G, 6-31G, 6-31G (d), and 6-31G (d, p)] and inserted them into the database. After selecting the basis set, the hybrid molecular orbitals block for a specific basis set can be automatically written into the $FMOHYB group of the input file.
Once generated, the QM input files for MFCC, or GAMESS for FMO, can be submitted to the grid environment, where a ‘best’ destination node (according to the scheduling policy) is selected to finish the job. The input files can be processed concurrently, with each activity accelerated by parallel computing.