Quantum biochemical database: Interfacing quantum chemistry with biochemistry.

摘要

Few would dispute that quantum mechanics (QM) has played a pivotal role in our understanding of small molecules throughout the many fields chemistry, however it is only recently that linear scaling algorithms combined with faster computers has made QM a viable method in the study of biological, macromolecules as well. At the same time, an explosion of available databases, both on the World Wide Web and from proprietary sources, have provided biochemists with an ever-deepening understanding of biochemical structure and function. The Quantum Biochemical Database (QBioDB) is an amalgamation of both of these significant trends, as it is a relational database consisting of the quantum mechanical description of thousands of macromoleculer structures including proteins, DNA, and RNA. This portable database, currently built on the MySQL Relational Database Management System, has been normalized to 25 key relations encompassing the three-dimensional, fully-protonated structures of over 5000 high-resolution biomolecules from the Protein Data Bank. In addition to structural data, these relations include attributes describing the quantum mechanically-derived atomic charges based on three models (CM1, CM2, and Muliken); frontier molecular orbital eigenvectors; active site data as defined in the PDB; fully protonated HETGROUPS along with characteristic Generalized AMBER Force Field (GAFF) atom types; overall structural energies; and others. To date, the AM1 and PM3 semiempirical Hamiltonians have been employed for all quantum chemical characterizations; however, higher levels of theory could be utilized in the future as they become more feasible, and the QBioDB has been built with this extensibility in mind. In addition to the data structures themselves, the QBioDB Toolkit has also been developed to encapsulate the many steps required in the construction and population of the QBioDB along with the characterization of the biomolecules it contains. With this Toolkit, it is expected that the QBioDB will be updated regularly as the PDB is updated. In a first "model analysis" of the data within the QBioDB, the frontier molecular orbital (FMO) localization and energies of the entire population of enzymes was performed. It was found that, when including both the active site and the enclosed ligand, the highest occupied molecular orbital (HOMO) was localized on these areas a full 87% in the solvated case as compared to 32% in the unsolvated case. Similarly, in the same population, the lowest occupied molecular orbital (LUMO) was localized on these atoms almost 100% of the time in the solvated case as opposed to 50% of the time in the unsolvated case. Together, these results support the notion that the most reactive orbitals in the structure, as defined by FMO theory, are generally found at those places in the structure where there is quantifiable evidence of biological activity (i.e., in the active site). This discovery could be used in the future to aid in our understanding of newly characterized enzymatic structures by providing us with a mechanism to find active regions of the biomolecule. Additional analyses will certainly be performed using the QBioDB in the future, and whether one studies the quantum chemical characteristics of a single biological macromolecule, or of the entire population, this database will help in the general acceptance and use of quantum chemical theory in biochemistry. It is hoped that as the database becomes more useful, that it will become an integral tool within the computational toolbox of the modern biochemist.