The protein kinases are a large family of enzymes that play a fundamental role in propagating signals within the cell. Because of the high degree of binding site similarity shared among protein kinases, designing drug compounds with high specificity among the kinases has proven difficult. However, computational approaches to comparing the 3-dimensional geometry and physicochemical properties of key binding site residues, referred to here as substructures, have been shown to be informative of inhibitor selectivity. This thesis introduces two fundamental approaches for the comparative analysis of substructure similarity and demonstrates the importance of each method on a variety of large protein structure datasets for multiple biological applications. The Family-wise Alignment of SubStructural Templates Framework (The FASST Framework) provides an unsupervised learning approach for identifying substructure clusterings. The substructure clusterings identified by FASST allow for the automatic evaluation of substructure variability, the identification of distinct structural conformations and the selection of anomalous outlier structures within large structure datasets. These clusterings are shown to be capable of identifying biologically meaningful structure trends among a diverse number of protein families. The FASST Live visualization and analysis platform provides multiple comparative analysis pipelines and allows the user to interactively explore the substructure clusterings computed by FASST. The Combinatorial Clustering Of Residue Position Subsets (CCORPS) method provides a supervised learning approach for identifying structural features that are correlated with a given set of annotation labels. The ability of CCORPS to identify structural features predictive of functional divergence among families of homologous enzymes is demonstrated across 48 distinct protein families. The CCORPS method is further demonstrated to generalize to the very difficult problem of predicting protein kinase inhibitor affinity. CCORPS is demonstrated to make perfect or near-perfect predictions for the binding ability of 12 of the 38 kinase inhibitors studied, while only having overall poor predictive ability for 1 of the 38 compounds. Additionally, CCORPS is shown to identify shared structural features across phylogenetically diverse groups of kinases that are correlated with binding affinity for particular inhibitors; such instances of structural similarity among phylogenetically diverse kinases are also shown to not be rare among kinases. Finally, these function-specific structural features may serve as potential starting points for the development of highly specific kinase inhibitors. Importantly, both The FASST Framework and CCORPS implement a redundancy-aware approach to dealing with structure overrepresentation that allows for the incorporation of all available structure data. As shown in this thesis, surprising structural variability exists even among structure datasets consisting of a single protein sequence. By incorporating the full variety of structural conformations within the analysis, the methods presented here provide a richer view of the variability of large protein structure datasets.

A drug is typically a small molecule that interacts with the binding site of some target protein. Drug design involves the optimization of this interaction so that the drug effectively binds with the target protein while not binding with other proteins (an event that could produce dangerous side effects). Computational drug design involves the geometric modeling of drug molecules, with the goal of generating similar molecules that will be more effective drug candidates. It is necessary that algorithms incorporate strategies to measure molecular similarity by comparing molecular descriptors that may involve dozens to hundreds of attributes. We use kernel-based methods to define these measures of similarity. Kernels are general functions that can be used to formulate similarity comparisons. The overall goal of this thesis is to develop effective and efficient computational methods that are reliant on transparent mathematical descriptors of molecules with applications to affinity prediction, detection of multiple binding modes, and generation of new drug leads. While in this thesis we derive computational strategies for the discovery of new drug leads, our approach differs from the traditional ligandbased approach. We have developed novel procedures to calculate inverse mappings and subsequently recover the structure of a potential drug lead. The contributions of this thesis are the following: 1. We propose a vector space model molecular descriptor (VSMMD) based on a vector space model that is suitable for kernel studies in QSAR modeling. Our experiments have provided convincing comparative empirical evidence that our descriptor formulation in conjunction with kernel based regression algorithms can provide sufficient discrimination to predict various biological activities of a molecule with reasonable accuracy. 2. We present a new component selection algorithm KACS (Kernel Alignment Component Selection) based on kernel alignment for a QSAR study. Kernel alignment has been developed as a measure of similarity between two kernel functions. In our algorithm, we refine kernel alignment as an evaluation tool, using recursive component elimination to eventually select the most important components for classification. We have demonstrated empirically and proven theoretically that our algorithm works well for finding the most important components in different QSAR data sets. 3. We extend the VSMMD in conjunction with a kernel based clustering algorithm to the prediction of multiple binding modes, a challenging area of research that has been previously studied by means of time consuming docking simulations. The results reported in this study provide strong empirical evidence that our strategy has enough resolving power to distinguish multiple binding modes through the use of a standard k-means algorithm. 4. We develop a set of reverse engineering strategies for QSAR modeling based on our VSMMD. These strategies include: (a) The use of a kernel feature space algorithm to design or modify descriptor image points in a feature space. (b) The deployment of a pre-image algorithm to map the newly defined descriptor image points in the feature space back to the input space of the descriptors. (c) The design of a probabilistic strategy to convert new descriptors to meaningful chemical graph templates. The most important aspect of these contributions is the presentation of strategies that actually generate the structure of a new drug candidate. While the training set is still used to generate a new image point in the feature space, the reverse engineering strategies just described allows us to develop a new drug candidate that is independent of issues related to probability distribution constraints placed on test set molecules.

The main topic of this thesis is the development of computational search methods that are useful in drug design applications. The emphasis is on exhaustiveness of the search method such that it can guarantee a certain level of geometric accuracy. In particular, the following two problems are addressed: (i) Prediction of binding mode of a drug molecule to a receptor and (ii) prediction of crystal structures of drug molecules. Predicting the binding mode(s) of a drug molecule to a target receptor is pivotal in structure-based rational drug design. In contrast to most approaches to solve this problem, the idea in this work is to analyze the search problem from a computational perspective. By building on top of an existing docking tool, new methods are proposed and relevant computational results are proven. These methods and results are applicable for other place-and-join frameworks as well. A fast approximation scheme for the docking of rigid fragments is described that guarantees certain geometric approximation factors. It is also demonstrated that this can be translated into an energy approximation for simple scoring functions. A polynomial time algorithm is developed for the matching phase of the docked rigid fragments. It is demonstrated that the generic matching problem is NP-hard. At the same time the optimality of the proposed algorithm is proven under certain scoring function conditions. The matching results are also applicable for some of the fragment-based de novo design methods. On the practical side, the proposed method is tested on 829 complexes from the PDB. The results show that the closest predicted pose to the native structure has the average RMS deviation of 1.06 °A. The prediction of crystal structures of small organic molecules has significantly improved over the last two decades. Most of the new developments, since the first blind test held in 1999, have occurred in the lattice energy estimation subproblem. In this work, a new efficient systematic search method that avoids random moves is proposed. It systematically searches through the space of possible crystal structures and conducts search space cuts based on statistics collected from the structural databases. It is demonstrated that the fast search method for rigid molecules can be extended to include flexible molecules as well. Also, the results of some prediction experiments are provided showing that in most cases the systematic search generates a structure with less than 1.0°A RMSD from the experimental crystal structure. The scoring function that has been developed for these experiments is described briefly. It is also demonstrated that with a more accurate lattice energy estimation function, better results can be achieved with the proposed robust search method.

A drug is typically a small molecule that interacts with the binding site of some target protein. Drug design involves the optimization of this interaction so that the drug effectively binds with the target protein while not binding with other proteins (an event that could produce dangerous side effects). Computational drug design involves the geometric modeling of drug molecules, with the goal of generating similar molecules that will be more effective drug candidates. It is necessary that algorithms incorporate strategies to measure molecular similarity by comparing molecular descriptors that may involve dozens to hundreds of attributes. We use kernel-based methods to define these measures of similarity. Kernels are general functions that can be used to formulate similarity comparisons. The overall goal of this thesis is to develop effective and efficient computational methods that are reliant on transparent mathematical descriptors of molecules with applications to affinity prediction, detection of multiple binding modes, and generation of new drug leads. While in this thesis we derive computational strategies for the discovery of new drug leads, our approach differs from the traditional ligandbased approach. We have developed novel procedures to calculate inverse mappings and subsequently recover the structure of a potential drug lead. The contributions of this thesis are the following: 1. We propose a vector space model molecular descriptor (VSMMD) based on a vector space model that is suitable for kernel studies in QSAR modeling. Our experiments have provided convincing comparative empirical evidence that our descriptor formulation in conjunction with kernel based regression algorithms can provide sufficient discrimination to predict various biological activities of a molecule with reasonable accuracy. 2. We present a new component selection algorithm KACS (Kernel Alignment Component Selection) based on kernel alignment for a QSAR study. Kernel alignment has been developed as a measure of similarity between two kernel functions. In our algorithm, we refine kernel alignment as an evaluation tool, using recursive component elimination to eventually select the most important components for classification. We have demonstrated empirically and proven theoretically that our algorithm works well for finding the most important components in different QSAR data sets. 3. We extend the VSMMD in conjunction with a kernel based clustering algorithm to the prediction of multiple binding modes, a challenging area of research that has been previously studied by means of time consuming docking simulations. The results reported in this study provide strong empirical evidence that our strategy has enough resolving power to distinguish multiple binding modes through the use of a standard k-means algorithm. 4. We develop a set of reverse engineering strategies for QSAR modeling based on our VSMMD. These strategies include: (a) The use of a kernel feature space algorithm to design or modify descriptor image points in a feature space. (b) The deployment of a pre-image algorithm to map the newly defined descriptor image points in the feature space back to the input space of the descriptors. (c) The design of a probabilistic strategy to convert new descriptors to meaningful chemical graph templates. The most important aspect of these contributions is the presentation of strategies that actually generate the structure of a new drug candidate. While the training set is still used to generate a new image point in the feature space, the reverse engineering strategies just described allows us to develop a new drug candidate that is independent of issues related to probability distribution constraints placed on test set molecules.

The main topic of this thesis is the development of computational search methods that are useful in drug design applications. The emphasis is on exhaustiveness of the search method such that it can guarantee a certain level of geometric accuracy. In particular, the following two problems are addressed: (i) Prediction of binding mode of a drug molecule to a receptor and (ii) prediction of crystal structures of drug molecules. Predicting the binding mode(s) of a drug molecule to a target receptor is pivotal in structure-based rational drug design. In contrast to most approaches to solve this problem, the idea in this work is to analyze the search problem from a computational perspective. By building on top of an existing docking tool, new methods are proposed and relevant computational results are proven. These methods and results are applicable for other place-and-join frameworks as well. A fast approximation scheme for the docking of rigid fragments is described that guarantees certain geometric approximation factors. It is also demonstrated that this can be translated into an energy approximation for simple scoring functions. A polynomial time algorithm is developed for the matching phase of the docked rigid fragments. It is demonstrated that the generic matching problem is NP-hard. At the same time the optimality of the proposed algorithm is proven under certain scoring function conditions. The matching results are also applicable for some of the fragment-based de novo design methods. On the practical side, the proposed method is tested on 829 complexes from the PDB. The results show that the closest predicted pose to the native structure has the average RMS deviation of 1.06 °A. The prediction of crystal structures of small organic molecules has significantly improved over the last two decades. Most of the new developments, since the first blind test held in 1999, have occurred in the lattice energy estimation subproblem. In this work, a new efficient systematic search method that avoids random moves is proposed. It systematically searches through the space of possible crystal structures and conducts search space cuts based on statistics collected from the structural databases. It is demonstrated that the fast search method for rigid molecules can be extended to include flexible molecules as well. Also, the results of some prediction experiments are provided showing that in most cases the systematic search generates a structure with less than 1.0°A RMSD from the experimental crystal structure. The scoring function that has been developed for these experiments is described briefly. It is also demonstrated that with a more accurate lattice energy estimation function, better results can be achieved with the proposed robust search method.

Docking is one of the methods in virtual screeing. Studies from around 1980 to now, many docking software have been developed, but these software have many short comings. The software currently used for docking have many disadvantage: poor efficiency, rigid structure of the proteins and the ligands, poor accuracy, without the polarization after binding, leading virtual screening is still stuck in a supporting role. Our experiment with new method improves those shortcomings of docking. With this new method, we obtain the following improvements in docking process: better efficiency, flexible structure of the proteins and the ligands, better accuracy. In the depression-related protein docked with traditional Chinese medicine test. We change the conformations of ligands with the shapes of active sites before posing, this makes the conformation of complex much more reasonable, even more complicated, large ligands. In the experiment of random sites docking, we found a possible path for compounds traveling into active sites. We illustrate a docking area by linking all possible docking sites. The lead compound may not successfully travel into active site when this area is occupied by other proteins or ligands. In the docking experiment with side-chain rotation, we rotate the torsion angle to make side chains relax. We obtained a similar result with molecular dynamics, and saved a lot of time.

Computational methods are a convenient resource to solve drawbacks of drug research such as high cost, time-consumption, and high risk of failure. In order to get an optimum search of new drugs we need to design a rational approach to analyze the molecular forces that govern the interactions between the drugs and their target molecules. The objective of this project is to get an understanding of the interactions between drugs and proteins at the molecular level. The interaction energy, when protein and drugs react, has two components: non-covalent and covalent. The former accounts for the ionic interactions, the later accounts for electron transfer between the reactants. We study each energy component using the most popular analysis tools in computational chemistry such as docking scoring, molecular dynamics fluctuations, electron density change, molecular electrostatic potential (MEP), density of states projections, and the transmission function. We propose the probability of transfer of electrons (transmission function) between reactants in protein-drug complexes as an alternative tool for molecular recognition and as a direct correlator to the binding affinity. The quadratic correlation that exists between the electron transfer rate and the electronic coupling strength of the reactants allow a clear distinguishability between ligands. Thus, in order to analyze the binding affinity between the reactants, a calculation of the electronic coupling between them is more suitable than an overall energetic analysis such as free reaction energy.

Thesis (Ph. D.)--University of Washington, 2000. / Vita. Includes bibliographical references (leaves 175-193).

Ricin is a potent cytotoxin easily purified in large quantities. It presents a significant public health concern due to its potential use as a bioterrorism agent. For this reason, extensive efforts have been underway to develop antidotes against this deadly poison. The catalytic A subunit of the heterodimeric toxin has been biochemically and structurally well characterized, and is an attractive target for structure-based drug design. Aided by computer docking simulations, several ricin toxin A chain (RTA) inhibitors have been identified; the most promising leads belonging to the pterin family. To date, the most potent RTA inhibitors developed using this approach are only modest inhibitors with apparent IC50 values in the 10-4 M range, leaving significant room for improvement. This thesis discusses the development of a subset of inhibitors belonging to the pterin family in which amino acids have been utilized as building blocks. Inhibitors in this family have achieved a significant increase in potency, and have provided valuable structural information for further development. / text

Nitric oxide synthases (NOS) are responsible for the production of nitric oxide (NO), an essential cell-signaling molecule, in mammals. There are three isoforms of NOS with widely different tissue distribution. The overproduction of NO is marked in many human disease states and cancers, however due to the similarities of the enzyme isoforms, targeting NOS for inhibition has proven challenging. Endogenously, the methylated arginines, N[superscript ω]-monomethyl-L-arginine (NMMA) and asymmetric N[superscript ω], N[superscript ω]-dimethyl-L-arginine (ADMA), inhibit NOS. N[superscript ω], N[superscript ω]-Dimethylarginine dimethylaminohydrolase (DDAH1) metabolizes these methylated arginines and thus relieves NOS inhibition. The role of DDAH1 in the regulation of diseases such as cancer and septic shock is still being elucidated. It is thought that targeting DDAH1 for inhibition rather than NOS may circumvent many of the current problems with the treatment of NO overproduction such as isoform selectivity. My PhD studies focus on the synthesis of a series of irreversible inhibitors of DDAH1, an extensive study of their in vitro mode of inhibition, a comparison of analytical fitting methods, and the viability and efficacy of the inactivators in a human cell line. I also studied a potential endogenous inactivator of DDAH1, nitroxyl (HNO), a one-electron reduction product of NO. / text