Computational methods for predicting the binding affinity of small organic molecules tobiological macromolecules cover a vast range of theoretical and physical complexity. Generally, as the required accuracy increases so does the computational cost, thereby making the user choose a method that suits their needs within the parameters of the project. We present how WScore, a rigid-receptor docking program normally consigned to structure-based hit discovery in drug design projects, is systematically ameliorated to perform accurately enough for lead optimization with a set of ROCK1 complexes and congeneric ligands from a structure-activity relationship study. Initial WScore results from the Schrödinger 2019-3 release show poor correlation (R² ∼0.0), large errors in predicted binding affinity (RMSE = 2.30 kcal/mol), and bad native pose prediction (two RMSD > 4Å) for the six ROCK1 crystal structures and associated active congeneric ligands. Improvements to WScore’s treatment of desolvation, myriad code fixes, and a simple ensemble consensus scoring protocol improved the correlation (R² = 0.613), the predicted affinity accuracy (RMSE = 1.34 kcal/mol), and native pose prediction (one RMSD > 1.5Å). Then we evaluate a physically and thermodynamically rigorous free energy perturbation (FEP) method, FEP+, against CryoEM structures of the Machilis hrabei olfactory receptor, MhOR5, and associated dose-response assays of a panel of small molecules with the wild-type and mutants. Augmented with an induced-fit docking method, IFD-MD, FEP+ performs well for ligand mutating relative binding FEP (RBFEP) calculations which correlate with experimental log(EC50)with an R² = 0.551. Ligand absolute binding FEP (ABFEP) on a set of disparate ligands from the MhOR5 panel has poor correlation (R² = 0.106) for ligands with log(EC50) within the assay range. But qualitative predictions correctly identify the ligands with the lowest potency. Protein mutation calculations have no log(EC50) correlation and consistently fail to predict the loss of potency for a majority of MhOR5 single point mutations. Prediction of ligand efficacy (the magnitude of receptor response) is also an unsolved problem as the canonical active and inactive conformations of the receptor are absent in the FEP simulations. We believe that structural insights of the mutants for both bound and unbound (apo) states are required to better understand the shortcomings of the current FEP+ methods for protein mutation RBFEP. Finally, improvements to GPU-accelerated linear algebra functions in an Auxiliary-Field Quantum Monte Carlo (AFQMC) program effect an average 50-fold reduction in GPU kernel compute time using optimized GPU library routines instead of custom made GPU kernels. Also MPI parallelization of the population control algorithm that destroys low-weight walkers has a bottleneck removed in large, multi-node AFQMC calculations.Computational methods for predicting the binding affinity of small organic molecules tobiological macromolecules cover a vast range of theoretical and physical complexity. Generally, as the required accuracy increases so does the computational cost, thereby making the user choose a method that suits their needs within the parameters of the project. We present how WScore, a rigid-receptor docking program normally consigned to structure-based hit discovery in drug design projects, is systematically ameliorated to perform accurately enough for lead optimization with a set of ROCK1 complexes and congeneric ligands from a structure-activity relationship study. Initial WScore results from the Schrödinger 2019-3 release show poor correlation (R² ∼0.0), large errors in predicted binding affinity (RMSE = 2.30 kcal/mol), and bad native pose prediction (two RMSD > 4Å) for the six ROCK1 crystal structures and associated active congeneric ligands. Improvements to WScore’s treatment of desolvation, myriad code fixes, and a simple ensemble consensus scoring protocol improved the correlation (R² = 0.613), the predicted affinity accuracy (RMSE = 1.34 kcal/mol), and native pose prediction (one RMSD > 1.5Å). Then we evaluate a physically and thermodynamically rigorous free energy perturbation (FEP) method, FEP+, against CryoEM structures of the Machilis hrabei olfactory receptor, MhOR5, and associated dose-response assays of a panel of small molecules with the wild-type and mutants. Augmented with an induced-fit docking method, IFD-MD, FEP+ performs well for ligand mutating relative binding FEP (RBFEP) calculations which correlate with experimental log(EC50)with an R² = 0.551. Ligand absolute binding FEP (ABFEP) on a set of disparate ligands from the MhOR5 panel has poor correlation (R² = 0.106) for ligands with log(EC50) within the assay range. But qualitative predictions correctly identify the ligands with the lowest potency. Protein mutation calculations have no log(EC50) correlation and consistently fail to predict the loss of potency for a majority of MhOR5 single point mutations. Prediction of ligand efficacy (the magnitude of receptor response) is also an unsolved problem as the canonical active and inactive conformations of the receptor are absent in the FEP simulations. We believe that structural insights of the mutants for both bound and unbound (apo) states are required to better understand the shortcomings of the current FEP+ methods for protein mutation RBFEP. Finally, improvements to GPU-accelerated linear algebra functions in an Auxiliary-Field Quantum Monte Carlo (AFQMC) program effect an average 50-fold reduction in GPU kernel compute time using optimized GPU library routines instead of custom made GPU kernels. Also MPI parallelization of the population control algorithm that destroys low-weight walkers has a bottleneck removed in large, multi-node AFQMC calculations.

Electron-driven processes lie at the core of a large variety of physical, biological, and chemical phenomena. Despite their crucial roles in science and technology, detailed description of these processes remains a significant challenge, and there is a need for the development of accurate and efficient computational tools that enable predictive simulation. This work is focused on the development of novel software tools and methodologies aimed at two classes of electron-mediated processes: (i) electron-molecule scattering, and (ii) charge transfer in proteins. The first major focus of this thesis is the electronic structure of autoionizing electronic resonances. The theoretical description of these metastable states is intractable by means of conventional quantum chemistry techniques, and specialized techniques are required in order to accurately describe their energies and lifetimes. In this work, we have utilized the complex absorbing potential (CAP) method, and describe three developments which have advanced the applicability, efficiency, and accessibility of the CAP methodology for molecular resonances: (1) implementation and investigation of the smooth Voronoi potential (2) implementation of CAP in the projected scheme, and (3) development of the OpenCAP package, which extends the CAP methodology to popular electronic structure packages. The second major focus is the identification of electron and hole transfer (ET) pathways in biomolecules. Both experimental and theoretical inquiries into electron/hole transfer processes in biomolecules generally require targeted approaches, which are complicated by the existence of numerous potential pathways. To this end, we have developed an open-source web platform, eMap, which exploits a coarse-grained model of the protein crystal structure to (1) enable pre-screening of potentially efficient ET pathways, and (2) identify shared pathways/motifs in families of proteins. Following introductory chapters on motivation and theoretical background, we devote a chapter to each new methodology mentioned above. The open-source software tools discussed herein are under active development, and have been utilized in published work by several unaffiliated experimental and theoretical groups across the world. We conclude the dissertation with a summary and discussion of the outlook and future directions of the OpenCAP and eMap software packages.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

The Diels-Alder is an important reaction that is one of the primary tools for synthesizing cyclic carbon structures, while simultaneously introducing up to four stereocenters in the resulting product. Not only is it a widely explored reaction in organic chemistry, but a vital tool in industry to construct novel compounds for pharmacological applications. Still, a remaining concern is the fact that upon the introduction of stereogenic carbons, the possibility of stereoselective control is greatly diminished. A common solution to the problem of undesirable stereoisomers is to employ chiral auxiliaries and ligands as means to increase the yield of a certain stereoisomer. However, incorporating these types of compounds in order to obtain an enantiomerically pure product increases the amount of synthetic steps to be regulated, implying that one or more purification steps are necessary to obtain the desired result. An accompanying thought leans toward the environmental aspect, as the principles of green chemistry are of great importance. This thesis presents the attempts to explore the possibility of engineering an enzyme that can catalyze an asymmetric Diels-Alder reaction through the use of molecular modeling. Based on previous work, the catalytically proficient enzyme ketosteroid isomerase had been deemed a probable candidate as a Diels-Alderase. To evaluate the enzyme thoroughly, a set of compounds was scored against the active binding site where the best hits against the wild type were saved and evaluated repeatedly after the introduction of rational mutations. Although no conclusive indication of an optimal design could be obtained at the end of this work, valuable insight was retrieved on plausible design strategies, which eventually could help lead to the first catalytically proficient Diels-Alderase. / Diels-Alder är en viktig reaktion då den är ett redskap för att syntetisera cykliska kolstrukturer, samtidigt som uppemot fyra stereocentra introduceras i den resulterande produkten. Reaktionen används inte enbart inom organisk kemi, utan är även ett viktigt redskap inom industriella sammanhang för att ta fram nya preparat som direkt kan tillämpas inom farmakologi. En återstående problematik är faktumet att introduktionen av nya stereogena kol bidrar till att drastiskt minska möjligheten att bibehålla en stereoselektiv kontroll. En vanlig lösning för att undvika oönskade stereoisomerer är att nyttja kirala hjälpmolekyler och ligander för att öka utbytet av en specifik stereoisomer. Dock innebär införandet av dessa hjälpmolekyler i strävan att erhålla en enantiomeriskt ren produkt ett ökat antal syntes-steg att hantera, vilket antyder att ett eller flera reningssteg är nödvändiga för att uppnå önskat resultat. Ur en miljösynpunkt är detta värt att ha i åtanke, då principerna för grön kemi är viktiga. Detta arbete utforskar möjligheterna att konstruera ett enzym som kan katalysera en asymmetrisk Diels-Alder-reaktion, med hjälp av molekylär modellering. Baserat på tidigare arbeten har enzymet ketosteroid isomeras valts ut som en potential kandidat till ett Diels-Alderase. För att noggrant evaluera enzymet så screenades ett set av substrat mot dess aktiva säte, där de bästa träffarna gentemot vildtypen sparades och återevaluerades allteftersom rationella mutationer kontinuerligt introducerades. Trots avsaknaden av klara indikationer på att en optimal design har kunnat tas fram vid slutet av detta arbete, så erhölls värdefull insikt på möjliga design-strategier, vilket skulle kunna bistå sökandet av det första katalytiskt effektiva Diels-Alderase.

Geometry optimization is a key step in the computational modeling of chemical reactions because one cannot model a chemical reaction without first accurately determining the molecular structure, and electronic energy, of the reactants and products, along with the transition state that connects them. These structures are stationary points— the reactant and product structures are local minima, and the transition state is a saddle point with one negative-curvature direction—on the molecular potential energy surface. Over the years, many methods for locating these stationary points have been developed. In general, the problem of finding reactant and product structures is relatively straightforward, and reliable methods exist. Converging to transition states is much more challenging. Because of the difficulty of transition-state optimization, researchers have designed optimization methods specifically for this problem. These methods try to make good choices for the initial geometry, the system of coordinates used to represent the molecule, the initial Hessian, the Hessian updating method, and the step-size. The transition-state optimization method developed in this thesis required considering all of these methods. Specifically, a new method for finding an initial guess geometry was developed in chapter 2; good choices for a coordinate system for representing the molecule were explored in chapters 2 and 6; different choices for the initial Hessian are considered in chapter 5; chapters 3 and 4 present, and test, a sophisticated new method for updating the Hessian and controlling the step-size during the optimization. iv The methods created in the process of this research led to the development of Saddle, a general-purpose geometry optimizer for transition states and stable structures, with and without constraints on the molecular coordinates. Saddle can be run in conjunction with the Gaussian program or almost any other quantum chemistry program, and it converges significantly more often than the other traditional methods we tested. / Thesis / Doctor of Science (PhD)