Selective Mu Opioid Receptor (MOR) antagonists possess immense potential in the treatment of opioid abuse/addiction. Utilizing the “message-address” concept, our laboratory reported a novel, reversible, non-peptide MOR selective antagonist 17-cyclopropylmethyl-3,14β-dihydroxy-4,5α-epoxy-6β-[(4՛-pyridyl)carboxamido]morphinan (NAP). Molecular modeling studies revealed that the selectivity of NAP for the MOR is because of a π-π stacking interaction of its pyridine ring with the Trp318residue in theMOR. Pharmacological characterization showed that NAP is a P-glycoprotein substrate, thereby limiting its use in the treatment of opioid abuse/addiction. Thus, to modify NAP, we replaced the pyridine ring with its isosteric counterpart thiophene. Isosteric replacement could lead to development of compounds with different pharmacologic properties. Additionally, exploring other ring systems would diversify and enrich our library of compounds and aid in establishing a comprehensive structure-activity relationship. Therefore, newly synthesized compounds included thiophene derivatives of 6α/β-naltrexamine with potential to be used in the treatment of opioid abuse/addiction. Preliminary in vivo screening revealed that compounds 8 and 11 could be acting as antagonists. To aid in the design and synthesis of newer generation of MOR selective analogs, a 3-Dimensional Quantitative Structure-Activity Relationship (3D-QSAR) Comparative Molecular Field Analysis (CoMFA) on 6β-N-heterocyclic substituted naltrexamine derivatives was conducted. After rigorous optimizations, the best CoMFA model possessed low predictive power. Results obtained suggested that small structural changes could lead to significant change in binding modes of these ligands. To further validate this observation, molecular docking studies were performed which revealed that these ligands indeed possessed multiple distinct binding modes thereby offering rationale for the CoMFA results. Thus, overall this study furnished useful information about the complexity of protein-ligand interactions which will aid in designing more potent and selective MOR ligands.

Freight distribution systems are under stress. With the world population growing, the migration of people to urban areas and technologies that allow purchases from virtually anywhere, efficient freight distribution can be challenging. An inefficient movement of goods may lead to business not being economically viable and also has social and environmental negative effects. An important strategy to be incorporated in freight distribution systems is the consolidation of goods, i.e., group goods by their destination. This strategy increases vehicles utilisation, reducing the number of vehicles and the number of trips required for the distribution and, consequently, costs, traffic, noise and air pollution. In this thesis, we explore consolidation in three different contexts (or cases) from an optimisation point of view. Each context is related to optimisation problems for which we developed mathematical programming models and solution methods. The first case in which we explore consolidation is in container loading problems (CLPs). CLPs are a class of packing problems which aims at positioning three-dimensional boxes inside a container efficiently. The literature has incorporated many practical aspects into container loading solution method (e.g. restricting orientation of boxes, stability and weight distribution). However, to the best of our knowledge, the case considering more dynamic systems (e.g. cross-docking) in which goods might have a schedule of arrival were yet to be contemplated by the literature. We define an extension of CLP which we call Container Loading Problem with Time Availability Constraints (CLPTAC), which considers boxes are not always available for loading. We propose an extension of a CLP model that is suitable for CLPTAC and solution methods which can also handle cases with uncertainty in the schedule of the arrival of the boxes. The second case is a more broad view of the network, considering an open vehicle routing problem with cross-dock selection. The traditional vehicle routing problem has been fairly studied. Its open version (i.e. with routes that start and end at different points) has not received the same attention. We propose a version of the open vehicle routing problem in which some nodes of the network are consolidation centres. Instead of shippers sending goods directly to their consumers, they must send to one of the available consolidation centres, then, goods are resorted and forwarded to their destination. For this problem, we propose a mixed integer linear programming model for cost minimisation and a solution method based on the Benders decomposition framework. A third case in which we explored consolidation is in collaborative logistics. Particularly, we focus on the shared use of the currently available infrastructure. We defined a hub selection problem in which one of the suppliers is selected as a hub. In a hub facility, other suppliers might meet to exchange their goods allowing one supplier to satisfy the demand from others. For this problem, we propose a mixed integer linear programming model and a heuristic based on the model. Moreover, we compared a traditional distribution strategy, with each supplier handling its demand, against the collaborative one. In this thesis, we explore these three cases which are related to consolidation for improving the efficiency in freight distribution systems. We extend some problems (e.g. versions of CLP) to apply them to a more dynamic setting and we also define optimisation problems for networks with consolidation centres. Furthermore, we propose solution methods for each of the defined problems and evaluate them using randomly generated instances, benchmarks from the literature and some cases based on real-world characteristics. / Sistemas de distribuição de carga possuem uma demanda muito alta. Com a população mundial crescendo, a migração em direção às áreas urbanas e as tecnologias que permitem compras de virtualmente qualquer lugar, a distribuição eficiente de mercadorias pode ser um desafio. Uma movimentação ineficiente de mercadorias pode tornar negócios economicamente inviáveis além de ter um impacto social e ambiental negativos. Uma estratégia importante para se incorporar em sistemas de distribuição é a consolidação de cargas, isto é, agrupar cargas de acordo com seus destinos. Essa estratégia aumenta a utilização dos veículos, reduzindo o número de veículos e viagens necessários para a distribuição e, consequentemente, custos, tráfego, poluição sonora e do ar. Nesta tese, é explorada a técnica de consolidação em três casos diferentes de um ponto de vista de otimização. Cada caso é relacionado a problemas de otimização para os quais são propostos modelos de programação matemática e métodos de solução. O primeiro caso em que é explorada a consolidação é em Problemas de Carregamento de Contêineres (PCCs). PCCs pertencem a uma classe de problemas de empacotamento que visa posicionar caixas tridimensionais dentro de contêineres eficientemente. A literatura tem incorporado diversos aspectos práticos em procedimentos de solução dos PCCs (por exemplo, restringir a orientação das caixas, estabilidade e distribuição de peso). No entanto, o caso que considera sistemas logísticos mais dinâmicos (como cross-docking), nos quais mercadorias podem ter uma agenda de chegada ainda não havia sido contemplados. É definida uma extensão de PCC chamada de Problema de Carregamento de Contêieneres com Restrições de Disponibilidade Temporal (PCCRDT). Também, propõem-se modelos e métodos de solução para o PCCRDT que são capazes de lidar com incerteza na chegada das mercadorias. O segundo caso utiliza uma visão mais abrangente da rede de distribuição, considerando um problema de roteamento de veículos em rede aberta com seleção de cross-dock. O problema tradicional de roteamento de veículos é bastante estudado. A sua versão aberta (com rotas que começam e terminam em pontos diferentes) não tem recebido tanta atenção. É proposta uma versão do roteamento de veículos em rede aberta em que alguns nós da rede são centros de consolidação. Os fornecedores, ao invés de enviar as mercadorias diretamente para os consumidores, enviam-nas para um dos centros de consolidação disponíveis, então, as mercadorias são reorganizadas (em diferentes veículos) e encaminhadas para o seus destinos. Para esse problema, é proposto um modelo de programação linear inteira mista para a minimização de custo e um método de solução baseado no arcabouço de decomposição de Benders. Um terceiro caso em que foi explorada a consolidação de mercadorias é o de logística colaborativa. Particularmente, se concentrou no uso compartilhado de infra-estrutura já disponível na rede de distribuição. É definido um problema de seleção de seleção de um dos fornecedores como hub. No hub, outros fornecedores podem se encontrar para trocar suas mercadorias, permitindo que um fornecedor satisfaça a demanda de outro. Para esse problema, é proposto um modelo de programação linear inteira mista e uma heurística baseada no modelo. Ainda, é comparada uma estratégia de distribuição convencional (com cada fornecedor responsável pela sua própria demanda) com uma estratégia colaborativa. Nesta tese, são explorados esses três casos que se relacionam com consolidação para melhorar a eficiência de sistemas de distribuição de carga. São estendidos alguns problemas (como o PCC) para que se possa aplicá-los em cenários mais dinâmicos e também são definidos problemas de otimização em redes com centros de consolidação. Além disso, são propostos métodos de solução para cada um dos casos. Os métodos são avaliados em instâncias geradas aleatoriamente, instâncias da literatura e, em alguns casos, instâncias baseadas em cenários reais.

For structure-based drug discovery, structural information of a target protein is necessary. NMR, or X-ray crystallography can provide necessary information on active site configuration that can lead a successful virtual screening campaign into identifying binders that may then be optimized into potent inhibitors. However, many challenges exist in the structure-based drug discovery cycle. For instance, structure determination of a protein of interest can many times be a daunting task. In addition, complex structure determination, which can allow essential characterization of protein-ligand interactions, is also challenging and many times impossible. Virtual screening heavily relies on such structural information, but hit-to-lead optimization schemes do as well. Furthermore, inherent protein characteristics such as conformational flexibility only add to the complexities in using structural information to identifying and optimizing inhibitors. In the scope of the work presented here, a structure-based drug discovery approach against three different protein targets is described. Each is presented with it's own set of challenges, but each has successfully led to the identification of new ligands. The drug discovery project against CXCL12 will first be described. CXCL12 is a small chemokine (~10KDa) that binds to the CXCR4 receptor promoting chemotaxis of lymphocytes but also metastasis of cancer cells. This interaction is further supported by sulfated tyrosines on CXCR4 that bind specific sites on the CXCL12 surface. The CXCL12-CXCR4 signaling axis has been a major focus of drug discovery, but efforts are mainly focused on CXCR4, since CXCL12 is a small protein lacking surface characteristics that are thought to be druggable. Yet, through a combination of rigid, flexible, and ensemble docking in virtual screening studies, we have successfully identified compounds that bind each of the three sulfotyrosine recognition sites on CXCL12, which normally bind the sulfated tyrosines on CXCR4 (sY7, sY12, and sY21). Furthermore, we have led a hit-to-lead approach in optimizing compounds against the sY21-binding site, aided by trivial information gained through crystallographic complex structure determination of CXCL12 bound by such a compound. We aim to eventually link compounds against different sites together and greatly improve potency. Next, the drug discovery project against P. aeruginosa LpxA will be described. In Gram-negative bacteria, the first step of lipid A biosynthesis is catalyzed by UDP-N-acetylglucosamine acyltrasferase (LpxA) through the transfer of a R-3-hydroxyacyl chain from the acyl carrier protein (ACP) to the 3'-hydroxyl group of UDP-GlcNAc. Acyl chain length selectivity varies between species of bacteria, but is highly specific and conserved within certain species. In E. coli and L. interrogans for example, LpxA is highly selective for longer R-3-hydroxyacil chains (C14 and C12 respectively), while in P. aeruginosa the enzyme is highly selective for R-3-hydroxydecanoyl, a 10-hydrocarbon long acyl chain. Three P. aeruginosa LpxA crystal structures will be described here for the first time; the apo form, the complex with its substrate UDP-GlcNAc, and the complex with its product UDP-3-O-(R-3-hydroxydecanoyl)-GlcNAc. A comparison between the APO form and complexes identifies key residues that position UDP-GlcNAc appropriately for catalysis, and supports the role of His121 in generating the nucleophile by interacting with the UDP-GlcNAc 3'-hydroxyl group. Furthermore, the product-complex structure supports the role of Met169 as the "hydrocarbon ruler", providing structural information on how P. aeruginosa LpxA is granted its exceptional selectivity for the 10-hydrocarbon long acyl chain. Structural information of the active site was subsequently used in designing virtual screening experiments that led to the identification of two ligands, confirmed by X-ray crystallography screening to bind to the active site. We aim to continue application of X-ray crystallography into screening compound binding, and to also use a hit-to-lead approach in compound optimization. Finally, the drug discovery project against the Tiam1 PDZ domain will be described. Tiam1 (T-cell lymphoma invasion and metastasis gene 1) is a GEF (guanine exchange factor) protein that activates Rac1 and initiates tumor formation. Tiam1 is regulated through its PDZ domain, which binds to syndecan1. We have successfully applied a virtual screening strategy to an existing crystallographic structure of the Tiam1 PDZ domain complexed to a syndecan1 peptide and identified four ligands that bind to the PDZ domain with low affinities. These compounds provide a starting point for future hit-to-lead optimization strategies.

Although the tremendous catalytic power of enzymes is widely recognized, their exact mechanisms of action are still a source of debate. In order to elucidate the origin of their power, it is necessary to look at individual residues and atoms, and establish their contribution to ligand binding, activation, and reaction. Given the present limitations of experimental techniques, only computational tools allow for such detailed analysis. During my PhD studies I have applied a variety of computational methods, reviewed in Chapter 2, to the study of two enzymes: DfrB dihydrofolate reductase (DHFR) and methyltetrahydrofolate: corrinoid/iron-sulfur protein methyltransferase (MeTr). ¶ The DfrB enzyme has intrigued microbiologists since it was discovered thirty years ago, because of its simple structure, enzymatic inefficiency, and its insensitivity to trimethoprim. This bacterial enzyme shows neither structural nor sequence similarity with its chromosomal counterpart, despite both catalysing the reduction of dihydrofolate (DHF) using NADPH as a cofactor. As numerous attempts to obtain experimental structures of an enzyme ternary complex have been unsuccessful, I combined docking studies and molecular dynamics simulations to produce a reliable model of the reactive DfrB•DHF•NADPH complex. These results, combined with published empirical data, showed that multiple binding modes of the ligands are possible within DfrB. ¶ Comprehensive sequence and structural analysis provided further insight into the DfrB family. The presence of the dfrB genes within integrons and their level of sequence conservation suggest that they are old structures that had been diverging well before the introduction of trimethoprim. Each monomer of the tetrameric active enzyme presents an SH3-fold domain; this is a eukaryotic auxiliary domain never found before as the sole domain of a protein, let alone as the catalytic one. Overall, DfrB DHFR seems to be a poorly adapted catalyst, a ‘minimalistic’ enzyme that promotes the reaction by facilitating the approach of the ligands rather than by using specific catalytic residues. ¶ MeTr initiates the Wood-Ljungdahl pathway of anaerobic CO2 fixation. It catalyses the transfer of the N5-methyl group from N5-methyltetrahydrofolate (CH3THF) to the cobalt centre of a corrinoid/iron-sulfur protein. For the reaction to occur, the N5 position of CH3THF is expected to be activated by protonation. As experimental studies have led to conflicting suggestions, computational approaches were used to address the activation mechanism. ¶ Initially, I tested the accuracy of quantum mechanical (QM) methods to predict protonation positions and pKas of pterin, folate, and their analogues. Then, different protonation states of CH3THF and active-site aspartic residues were analysed. Fragment QM calculations suggested that the pKa of N5 in CH3THF is likely to increase upon protein binding. Further, ONIOM calculations which accounted for the complete protein structure indicated that active-site aspartic residues are likely to be protonated before the ligand. Finally, solvation and binding free energies of several protonated forms of CH3THF were compared using the thermodynamic integration approach. Taken together, these preliminary results suggest that further work with particular emphasis on the protonation state of active-site aspartic residues is needed in order to elucidate the protonation and activation mechanism of CH3THF within MeTr.

Même si le docking protéine-protéine devient un outil incontournable pour répondre aux problématiques biologiques actuelles, il reste cependant deux difficultés inhérentes aux méthodes actuelles: 1) la majorité de ces méthodes ne considère pas les possibles déformations internes des protéines durant leur association. 2) Il n'est pas toujours simple de traduire les informations issues de la littérature ou d'expérimentations en contraintes intégrables aux programmes de docking. Nous avons donc tenté de développer une approche permettant d'améliorer les programmes de docking existants. Pour cela nous nous sommes inspirés des méthodologies mises en place sur des cas concrets traités durant cette thèse. D'abord, à travers la création du complexe ERBIN PDZ/Smad3 MH2, nous avons pu tester l'utilité de la Dynamique Moléculaire en Solvant Explicite (DMSE) pour mettre en évidence des résidus importants pour l'interaction. Puis, nous avons étendu cette recherche en utilisant divers serveurs de docking puis la DMSE pour cibler un résultat consensus. Enfin, nous avons essayé le raffinage par DMSE sur une cible du challenge CAPRI et comparé les résultats avec des simulations courtes de Monte-Carlo. La dernière partie de cette thèse portait sur le développement d'un nouvel outil de visualisation de la surface moléculaire. Ce programme, nommé MetaMol, permet de visualiser un nouveau type de surface moléculaire: la Skin Surface Moléculaire. La distribution des calculs à la fois sur le processeur de l'ordinateur (CPU) et sur ceux de la carte graphique (GPU) entraine une diminution des temps de calcul autorisant la visualisation, en temps réel, des déformations de la surface moléculaire.

Le cisplatine (ou cis-diammine, dichloro-platine) est l'un des composés chimiques les plus utilisés actuellement en chimiothérapie anticancéreuse. Depuis la description de ses propriétés anticancéreuses par B. Rosenberg en 1965 de nombreux travaux ont été effectués a n de décrire le mécanisme d'action lui conférant ses propriétés antitumorales. A travers une approche originale couplant modélisation et travail expérimental, les recherches réalisées durant ma thèse ont permis d'élucider le comportement dynamique d'un adduit platiné sur la séquence 5'-GCCG*G*GTCGC-3' / 5'-GCGACCCGGC-3' (G* représente une guanine platinée). Cette structure a été comparée à celle de l'adduit ADN-cisplatine déterminée précédemment au laboratoire sur la séquence G*G*A. Nous avons ainsi étudié l'influence d'une guanine adjacente en 3' au pontage GG-Pt sur la structure de l'adduit. Il s'agit de la première étude structurale sur un adduit du cisplatine avec la séquence GGG. Même si l'a nité de la séquence GGG et des sites contenant Gn (n>= 3) pour le platine (II) est connue depuis longtemps, ses adduits avec le cisplatine n'ont pas encore été étudiés par RMN à cause des problèmes posés par leur puri cation. Cette étude de l'adduit G*G*G par RMN a été confrontée à la description dynamique de cet adduit, calculée par simulation. La paramétrisation du champ de force parm 98 a été spéci quement a née pour mieux décrire l'environnement de l'atome du platine. La confrontation entre la simulation de l'adduit G*G*G-Pt et les données issues de l'étude par RMN a permis de valider notre paramétrisation. Pour déterminer les proportions précises des sous-états BI et BII de l'ADN, une méthode novatrice a été mise au point. Celle-ci est basée sur la combinaison de quatre distances inter-protons H2'(n)-H8(n), H1'(n)-H6/8(n+1), H2'(n)-H6/8(n+1) et H2''(n)-H6/8(n+1) qui permet de discriminer les deux sous-conformations. Ces améliorations du champ de force et de la méthode de détermination des sous-conformations BI/BII ont permis la description ne du comportement de l'adduit couplé à l'ADN, ce qui nous a servi pour étudier le mécanisme anti-tumoral du cisplatine. En effet la reconnaissance de l'ensemble cisplatine-ADN par une protéine (Lymhoïd Enhanced Factor I - LEF I) pourrait activer les voies métaboliques de la cellule cancéreuse conduisant vers l'apoptose ou vers la réparation de la tumeur. La simulation de l'ensemble ADN-protéine (sans cisplatine) a permis de présenter le mode de reconnaissance de la protéine sur la déformation ainsi que la mise en évidence de l'implication d'une molécule d'eau dans celui-ci. Les études sur le cisplatine fixé sur son ADN cible nous ayant apporté de nombreuses connaissances sur les déformations engendrées, nous avons validé la déformation structurale formée par un autre complexe de platine, le pyrazolato-bis-platine. Ce composé a été conçu de novo pour induire une déformation faible de son ADN cible a n de provoquer une réponse cellulaire différente de celle engendrée par le cisplatine. La simulation a indiqué que ce complexe induit une faible courbure de l'ADN et une déformation globale différente de celle du cisplatine, ce qui exclut le même mode de reconnaissance. Comme expérimentalement le complexe pyrazolato possède une activité anti-tumorale, les simulations effectuées suggèrent donc que le mode d'action au niveau cellulaire est différent du cisplatine. Les travaux réalisés lors de ma thèse ont permis d'améliorer la compréhension des déformations ADN-cisplatine, ADN-cisplatine-protéine et de perfectionner la description des composés platinés dans le champ de force parm 98 (puis parm 99) du logiciel de modélisation moléculaire AMBER.

<p>Structural bioinformatics deals with the analysis, classification and prediction of three-dimensional structures of biomacromolecules. It is becoming increasingly important as the number of structures is growing rapidly. This thesis describes three studies concerned with protein-function prediction and two studies about protein structure validation.</p><p>New protein structures are often compared to known structures to find out if they have a known fold, which may provide hints about their function. The functionality and performance of eleven fold-comparison servers were evaluated. None of the tested servers achieved perfect recall, so in practise a combination of servers should be used.</p><p>If fold comparison does not provide any hints about the function of a protein, structural motif searches can be employed. A survey of left-handed helices in known protein structures was carried out. The results show that left-handed helices are rare motifs, but most of them occur in active or ligand-binding sites. Their identification can therefore help to pinpoint potentially important residues.</p><p>Sometimes all available methods fail to provide hints about the function of a protein. Therefore, the potential of using docking techniques to predict which ligands are likely to bind to a particular protein has been investigated. Initial results show that it will be difficult to build a reliable automated docking protocol that will suit all proteins.</p><p>The effect of various phenomena on the precision of accessible surface area calculations was also investigated. The results suggest that it is prudent to report such values with a precision of 50 to 100 Ų.</p><p>Finally, a survey of register shifts in known protein structures was carried out. The identified potential register shifts were analysed and classified. A machine-learning approach ("rough sets") was used in an attempt to diagnose register errors in structures.</p>

Structural bioinformatics deals with the analysis, classification and prediction of three-dimensional structures of biomacromolecules. It is becoming increasingly important as the number of structures is growing rapidly. This thesis describes three studies concerned with protein-function prediction and two studies about protein structure validation. New protein structures are often compared to known structures to find out if they have a known fold, which may provide hints about their function. The functionality and performance of eleven fold-comparison servers were evaluated. None of the tested servers achieved perfect recall, so in practise a combination of servers should be used. If fold comparison does not provide any hints about the function of a protein, structural motif searches can be employed. A survey of left-handed helices in known protein structures was carried out. The results show that left-handed helices are rare motifs, but most of them occur in active or ligand-binding sites. Their identification can therefore help to pinpoint potentially important residues. Sometimes all available methods fail to provide hints about the function of a protein. Therefore, the potential of using docking techniques to predict which ligands are likely to bind to a particular protein has been investigated. Initial results show that it will be difficult to build a reliable automated docking protocol that will suit all proteins. The effect of various phenomena on the precision of accessible surface area calculations was also investigated. The results suggest that it is prudent to report such values with a precision of 50 to 100 Å2. Finally, a survey of register shifts in known protein structures was carried out. The identified potential register shifts were analysed and classified. A machine-learning approach ("rough sets") was used in an attempt to diagnose register errors in structures.

The hepatitis C virus (HCV), with a global prevalence of roughly 2%, is among the most serious diseases today. Among the more promising HCV targets is the NS3 protease, for which several drug candidates have entered clinical trials. In this work, computational methods have been developed and applied to the design of inhibitors of the HCV NS3 protease. Comparative molecular field analysis (CoMFA) modeling and molecular docking are the two main computational tools used in this work. CoMFA is currently the most widely used 3D-QSAR method. Methodology for improving its predictive performance by evaluating 6120 combinations of non-default parameters has been developed. This methodology was tested on 9 data sets for various targets and found to consistently provide models of enhanced predictive accuracy. Validation was performed using q2, r2pred and response variable randomization. Molecular docking was used to develop SARs in two series of inhibitors of the HCV NS3 protease. In the first series, preliminary investigations indicated that replacement of P2 proline with phenylglycine would improve potency. Docking suggested that phenylglycine-based inhibitors may participate in two additional interactions but that the larger, more flexible phenylglycine group may result in worse ligand fit, explaining the loss in potency. In the second series, β-amino acids were explored as α-amino acid substitutes. Although β-amino acid substitution may reduce the negative attributes of peptide-like compounds, this study showed that β-amino acid substitution resulted in reduced potency. The P3 position was least sensitive to substitution and the study highlighted the importance of interactions in the oxyanion hole. Finally, docking was used to provide the conformations and alignment necessary for a CoMFA model. This CoMFA model, derived using default settings, had q2 = 0.31 and r2pred = 0.56. Application of the optimization methodology provided a more predictive model with q2 = 0.48 and r2pred = 0.68.

Predicting the structure of complexes formed by two interacting proteins is an important problem in computation structural biology. Proteins perform many of their functions by binding to other proteins. The structure of protein-protein complexes provides atomic details about protein function and biochemical pathways, and can help in designing drugs that inhibit binding. Docking computationally models the structure of protein-protein complexes, given three-dimensional structures of the individual chains. Protein docking methods have two phases. In the first phase, a comprehensive, coarse search is performed for optimally docked models. In the second refinement and reranking phase, the models from the first phase are refined and reranked, with the expectation of extracting a small set of accurate models from the pool of thousands of models obtained from the first phase. In this thesis, new algorithms are developed for the refinement and reranking phase of docking. New scoring functions, or potentials, that rank models are developed. These potentials are learnt using large-scale machine learning methods based on mathematical programming. The procedure for learning these potentials involves examining hundreds of thousands of correct and incorrect models. In this thesis, hierarchical constraints were introduced into the learning algorithm. First, an atomic potential was developed using this learning procedure. A refinement procedure involving side-chain remodeling and conjugate gradient-based minimization was introduced. The refinement procedure combined with the atomic potential was shown to improve docking accuracy significantly. Second, a hydrogen bond potential, was developed. Molecular dynamics-based sampling combined with the hydrogen bond potential improved docking predictions. Third, mathematical programming compared favorably to SVMs and neural networks in terms of accuracy, training and test time for the task of designing potentials to rank docking models. The methods described in this thesis are implemented in the docking package DOCK/PIERR. DOCK/PIERR was shown to be among the best automated docking methods in community wide assessments. Finally, DOCK/PIERR was extended to predict membrane protein complexes. A membrane-based score was added to the reranking phase, and shown to improve the accuracy of docking. This docking algorithm for membrane proteins was used to study the dimers of amyloid precursor protein, implicated in Alzheimer's disease.R. DOCK/PIERR was shown to be among the best automated docking methods in community wide assessments. Finally, DOCK/PIERR was extended to predict membrane protein complexes. A membrane-based score was added to the reranking phase, and shown to improve the accuracy of docking. This docking algorithm for membrane proteins was used to study the dimers of amyloid precursor protein, implicated in Alzheimer’s disease. / text