Subsystem Quantum Mechanics and In Silico Medicinal and Biological Chemistry

Abstract

EditorialIn silico (or computational) biochemistry is a growing subfield of medicinal and biological chemistry. The growth of research in this area has accelerated at a dramatic rate during the past two decades given the phenomenal increase in the speed of computing and the drastic reduction of its costs. These technological advances were paralleled by no less spectacular conceptual advances in theoretical chemistry, bringing highquality electronic structure calculations, the full interpretative power of quantum mechanics, and computational speed together at the desktop of the practicing chemist. The pioneering early work of the Pullmans in the sixties has now come a very long way in modeling, interpreting and predicting the chemistry of life [1,2]. This relatively new Journal has already devoted a series of three special issues this year to computational medicinal chemistry, an initiative that underscores the importance of this emerging field in medicinal chemistry [3][4][5].The developments alluded to above may be classified into two broad categories: theories and methods that yield results of comparable quality to post-Hartree-Fock electronic structure methods, such as configuration interaction methods, but at a small fraction of the computational cost; and quantum mechanical theoretical advances that provide insight into the results of the calculations.A prime example of a new theory that often yields configuration interaction quality results at a much lower cost is density functional theory (DFT) [6,7], developed by Nobel Laureate Walter Kohn in the 1960s and that is now a widely used mature field.The advances in electronic structure methods were both theoretical (e.g., the development of new hybrid functionals, a noted example being the B3LYP [8,9] hybrid DFT functional) and also algorithmic by incorporating them into standard quantum mechanical codes such as GAMESS and Gaussian. DFT brought the cost of highquality calculations that incorporate Coulombic electron correlation to the fingertips of the average chemist, when the use of post-Hartree-Fock methods is impractical. But these advances were not all. Given that medicinal and biological chemists are typically interested in very large molecules, several innovations attacked the problem from a different angle: that of slowing down the scaling of the calculations with size.A number of methods were developed that focus on a region of the molecule such as an active site to study, for example, the catalysis of a biochemical reaction at this site or to compute and compare the binding energies of several substrates. These methods generally partition the system in layers surrounding the active site, with subsequent layers (subsystems) treated at more economical computational levels, with the active site being treated at the most accurate level and the layer most distant from it at the lowest level. A widely used example of this multilayered approach is the quantum mechanics/molecular mechanics approach [10][11][12][13][14].In other instances, the entire molecule...

Improving the Accuracy of Predicting protein–ligand binding-free Energy With Semiempirical Quantum Chemistry Charge

“…In addition, molecular fractionation with conjugate caps method proposed by Zhang et al could accurately calculate the interaction between protein and another molecule, which took advantages from decomposing protein molecule into amino acid-based fragments [34,35]. Other well-known fragmentation soft-scaling methods, such as kernel energy method, have also been proved powerful for predicting G of large biological molecules [36][37][38][39][40]. Within this context, we previously developed a method that utilized QM/MM method to calculate the partial atomic charge of protein-ligand complex, named QMPC, which could significantly improve the correlation between computationally predicted and experimental G [41].…”

Localization-Delocalization Matrices and Electron Density-Weighted Adjacency Matrices: New Electronic Fingerprinting Tools for Medicinal Computational Chemistry

“…It has recently been emphasized that extracting molecular descriptors from measurable 3D scalar fields such as the electron density [ρ(r)] and the electrostatic potential [V(r)], both fundamentally and uniquely related to the quantum state of the system, can yield in addition to robust QSAR models invaluable insight into the mode of action of drugs and their interactions with their receptors [13][14][15][16][17].…”

“…The electron density exhibits a characteristic topography that engenders a natural partitioning of the molecular 3D space into nonoverlapping atomic basins according to Bader's quantum theory of atoms in molecules [14][15][18][19][20]. This partitioning leads to the possibility of defining bond paths, a definition that brings concepts such as chemical bonding and structural stability into coincidence within real 3D space.…”