The efficiency of parallel tempering Monte Carlo is studied for a two-dimensional Ising system of length L with N=L^2 spins. An external field is used to introduce a difference in free energy between the two low temperature states. It is found that the number of replicas R_opt that optimizes the parallel tempering algorithm scales as the square root of the system size N. For two symmetric low temperature states, the time needed for equilibration is observed to grow as L^2.18. If a significant difference in free energy is present between the two states, this changes to L^1.02. It is therefore established that parallel tempering is sped up by a factor of roughly L if an asymmetry is introduced between the low temperature states. This confirms previously made predictions for the efficiency of parallel tempering. These findings should be especially relevant when using parallel tempering for systems like spin glasses, where no information about the degeneracy of low temperature states is available prior to the simulation.

This is a problem of sampling. The number of classical states of an N-body system grows with O( 3 ^ N ). To sample this space, advanced techniques are required. Replica Exchange (RE), also known as parallel tempering, is an example that uses parallelization, and Hamiltonian Replica Exchange is a subset of RE that scales the energy of the replicas. The number of simulations required grows at O( N^(1/2) ), where N is number of atoms in the system. Replica Exchange with Dynamical Scaling (REDS) attempts to address this problem to decrease computational cost. It has been shown to increase efficiency 10-fold. We implemented REDS in GROMACS 2018. (Abraham 2015) All changes to the source code were written in the form of parallel methods. Scripts were written in Python and Perl to automate the experiment entirely. An exchange connects a region of high energy space, far above the surface of the landscape, to low energy space, which approaches the surface of the landscape, which represents the natural conformational progression of the molecule. Using REDS we were able to achieve exchanges at temperatures spaced too far apart to exchange using normal RE. Ergo, the flexibility of dynamical scaling allowed regions of phase space that would have gone unsampled to be mapped, addressing our initial problem of sampling.

Polymers that undergo dramatic changes in structural conformations in response to numerous stimuli such as temperature, pH, electric and magnetic fields, light inten- sity, biological molecules, and solvent polarity, are known as stimuli-responsive or ”smart” polymers. There is a broad range of very promising applications of these materials in catalysis, environmental remediation, sensors or actuator systems, and as delivery systems of therapeutic agents. Researchers have been trying to mimic smart polymers based on properties of polymers found in nature such as proteins, carbohydrates and nucleic acids. Novel bio-compatible polymers with a variety of chemical functional groups, diverse topologies, and cross-linking patterns with the ability to self-assemble in vivo are being engineered. Experimental and theoretical studies indicate that the thermodynamic properties relating to the hydrophobic effects play a pivotal role in determining the self-assembly process in smart polymers. At the same time, computational approaches based on simulation and modeling provide an understanding of this phenomenon on the micro- scopic level. Building empirical models based on statistical mechanics methods and simulation data helps to design polymeric materials with desirable traits. My research is mainly focused on investigating physicochemical characteristics of stimuli-responsive polymers under different conditions. I used atomistic molecular dynamics simulations to investigate these effects on polymer conformation. Given the size and complexity of our polymeric systems, we employed Graphical Process- ing Units (GPU) and enhanced sampling techniques such as REDS2 to increase the sampling time. These methods allow for the study of polymeric structural dynamics in solvents of varying polarity and in human skin epidermis. Our constant pH simulation of poly(methacrylic acid) revealed that the overall response is made up of local and global structural changes. The local structural re- sponse depends on the tacticity of the polymer, which leads to distinct cooperative effects for polymers with varying stereochemistry. Such simulations help to under- stand the principal driving forces behind the mechanism of self-assembly processes.

The replica exchange method (REM) has been widely used in the computer simulation of complex systems, such as proteins, glasses, and atomic clusters, where conventional methods based on sampling the canonical ensemble struggle to attain ergodicity over a rugged energy landscape characterized by multiple minima separated by high energy barriers. While the standard temperature REM (tREM) has proven to be highly effective in the equilibrium sampling of stable single phase states, tREM is seriously challenged in the vicinity of a first-order phase transition. The generalized Replica Exchange Method (gREM) was developed to address these outstanding computational problems and provide a method for simulating strong phase transitions in condensed matter systems. The central idea behind gREM is to incorporate the merit of generalized ensemble sampling into the replica exchange paradigm. The key ingredients of gREM are parameterized effective sampling weights, which smoothly join ordered and disordered phases with a succession of unimodal energy distributions that transform unstable or metastable energy states of the canonical ensemble into stable states that can be fully characterized. The inverse mapping between the sampling weights and the effective temperature provides a sure way to design the effective sampling weights and achieve ergodic sampling. Various applications of gREM are presented, including studies of the solid-liquid phase change of an adapted Dzugutov model of glass formation, the mechanism of spinodal decomposition in the vapor-liquid transition of a simple fluid, and the apparent crossover from a first-order to continuous transition with increasing density in the freezing of a nanofilm of water confined between featureless and atomistic surfaces. Extensive gREM simulations combined with the Statistical Temperature Weighted Histogram Analysis Method (ST-WHAM) demonstrate the effectiveness of the approach and provide comprehensive characterization of thermodynamic and structural properties intrinsic to phase transitions in these diverse systems.

Submitted by Caroline Xavier (caroline.xavier@pucrs.br) on 2017-08-25T13:27:56Z No. of bitstreams: 1 TES_THIAGO_LIPINSKI_PAES_COMPLETO.pdf: 7285473 bytes, checksum: 4dab05982a53386f1d5dd2202a2f9b21 (MD5) / Made available in DSpace on 2017-08-25T13:27:57Z (GMT). No. of bitstreams: 1 TES_THIAGO_LIPINSKI_PAES_COMPLETO.pdf: 7285473 bytes, checksum: 4dab05982a53386f1d5dd2202a2f9b21 (MD5) Previous issue date: 2017-03-27 / Among the main computational techniques currently applied to study proteins, classical molecular dynamics plays a important hole, specially its variation called replica exchange molecular dynamics or REMD, which provides efficient conformational sampling. Regular secondary structures elements of proteins are formed and maintained via stabilization by hydrogen bonds within helices and between strands of a -sheet. Packing of these structural elements, allowed by flexible turns and loops connecting them, leads to the formation of a structure that, in the successful cases, represents the native, functional state of a protein. Ionic, dipole, van der Waals, hydrophobic interactions, and hydrogen bonding are fundamental to these events. Most of these forces are strong up to a distance of 4.0 ?. Hence, these are the distances involved in the formation of local structural nubs that can further propagate and form whole elements of secondary structure. The common practice while simulating is, however, to keep fixed the cutoff at values higher or equal to 8.0 ?. Here a novel replica exchange molecular dynamics approach based on running cutoffs (varying from 4.0 ? to 8.0 ?) to enhance protein structure prediction is presented. We first proved the method as a reproducible one, as well as following a Boltzmann distribution and sampling different structures of conventional REMD. The human villin headpiece protein (PDB ID: 1UNC) was used as case study. We tested 9 different simulation protocols, in triplicate, and proved the use of incremental cutoff as an effective approach to enhance the quality and speed of protein structure predictions via replica exchange molecular dynamics. Applying the method to the protein test set, although of limited size, CuT-REMD showed good performance against the ab initio methods, most of the time being either as the best prediction method or with close results to the best ones. This made it possible to also compare CuT-REMD with de novo methods. Despite the difficulties, CuT-REMD maintained a good performance even surpassing certain servers for all tested proteins. The results obtained are encouraging, with the emergence of new questions to be addressed in the future. / Dentre os principais m?todos computacionais aplicados atualmente ao estudo de prote?nas, a din?mica molecular cl?ssica realiza importante papel, especialmente sua varia??o intitulada Replica Exchange Molecular Dynamics ou REMD, a qual prov? amostragem conformacional eficiente. Elementos de Estruturas Secund?rias (EES) regulares de prote?nas s?o formados e mantidos atrav?s de estabiliza??o por liga??es de hidrog?nio dentro de h?lices e entre fitas de uma folha . O empacotamento desses elementos estruturais, permitido por voltas e la?os flex?veis conectando-os, leva ? forma??o de uma estrutura que, nos casos bem sucedidos, representa o estado nativo, funcional de uma prote?na. Intera??es i?nicas, dipolo-dipolo, de van der Waals e hidrof?bicas, al?m de liga??es de hidrog?nio, s?o fundamentais para esses eventos. A maioria dessas for?as ? mais forte at? uma dist?ncia de 4,0 ?. Assim, essas (de 0,0 ? a 4,0 ?) s?o as dist?ncias envolvidas na forma??o de estruturas locais, que podem ainda se propagar e formar elementos inteiros de estrutura secund?ria. A pr?tica comum ao se executar simula??es por DM ?, no entanto, manter um raio de corte fixo em valores maiores ou iguais a 8,0 ?. Esta tese apresenta o m?todo CuTREMD, uma nova abordagem de REMD com base em raio de corte incremental (variando de 4,0 ? a 8,0 ?) testando a hip?tese de que tal abordagem pode otimizar a predi??o de estruturas terci?rias de prote?nas. Primeiramente, foi utilizada a prote?na villin headpiece humana (c?digo PDB 1UNC), como estudo de caso, e nove diferentes protocolos de simula??o foram testados, todos em triplicata. Posteriormente, com base nos resultados obtidos, um protocolo-padr?o foi escolhido como protocolo CuT-REMD, e um conjunto de nove prote?nas adicionais foi testado, sendo os resultados comparados com o m?todo REMD convencional. A utiliza??o de raio de corte incremental provou-se uma abordagem eficaz para melhorar a qualidade e velocidade das predi??es de estruturas de prote?nas via REMD. Aplicando o m?todo ao conjunto teste de prote?nas, embora de tamanho limitado, CuT-REMD mostrou bom desempenho em rela??o aos m?todos ab initio, colocando-se na grande maioria das vezes ou como o melhor m?todo de predi??o ou com resultados pr?ximos aos melhores m?todos. Isso possibilitou compar?-lo tamb?m com m?todos de novo e, embora com mais dificuldade, CuT-REMD manteve bom desempenho, inclusive superando certos servidores em todas as ocasi?es. Os resultados obtidos, em suma, mostram-se encorajadores, com o surgimento de novos questionamentos a serem abordados futuramente.

Antibody microarrays constitute a next-generation sensing platform that has the potential to revolutionize the way that molecular detection is conducted in many scientific fields. Unfortunately, current technologies have not found mainstream use because of reliability problems that undermine trust in their results. Although several factors are involved, it is believed that undesirable protein interactions with the array surface are a fundamental source of problems where little detail about the molecular-level biophysics are known. A better understanding of antibody stability and antibody-antigen binding on the array surface is needed to improve microarray technology. Despite the availability of many laboratory methods for studying protein stability and binding, these methods either do not work when the protein is attached to a surface or they do not provide the atomistic structural information that is needed to better understand protein behavior on the surface. As a result, molecular simulation has emerged as the primary method for studying proteins on surfaces because it can provide metrics and views of atomistic structures and molecular motion. Using an advanced, coarse-grain, protein-surface model this study investigated how antibodies react to and function on different types of surfaces. Three topics were addressed: (1) the stability of individual antibodies on surfaces, (2) antibody binding to small antigens while on a surface, and (3) antibody binding to large antigens while on a surface. The results indicate that immobilizing antibodies or antibody fragments in an upright orientation on a hydrophilic surface can provide the molecules with thermal stability similar to their native aqueous stability, enhance antigen binding strength, and minimize the entropic cost of binding. Furthermore, the results indicate that it is more difficult for large antigens to approach the surface than small antigens, that multiple binding sites can aid antigen binding, and that antigen flexiblity simultaneously helps and hinders the binding process as it approaches the surface. The results provide hope that next-generation microarrays and other devices decorated with proteins can be improved through rational design.

La maladie d’Alzheimer touche plus de 40 millions de personnes dans le monde et résulte de l’agrégation du peptide beta-amyloïde de 40/42 résidus. En dépit de nombreuses études expérimentales et théoriques, le mécanisme de formation des fibres et des plaques n’est pas élucidé, et les structures des espèces les plus toxiques restent à déterminer. Dans cette thèse, je me suis intéressée à deux aspects. (1) La détermination du noyau de nucléation (N*) de deux fragments (Aβ)16-22 et (Aβ)37-42. Mon approche consiste à déterminer les paramètres OPEP du dimère (Aβ)16-22 en comparant des simulations Monte Carlo sur réseau et des dynamiques moléculaires atomiques par échange de répliques. Les paramètres fonctionnant aussi sur le trimère (Aβ)16-22 et les dimères et trimères (Aβ)37-42, j’ai étudié la surface d’énergie libre des décamères et mes simulations montrent que N* est de 10 chaines pour (Aβ)16-22 et est supérieure à 20 chaines pour (Aβ)37-42. (2) J’ai ensuite étudié les structures du dimère (Aβ)1-40 par simulations de dynamique moléculaire atomistique par échanges de répliques. Cette étude, qui fournit les conformations d’équilibre du dimère Aβ1-40 en solution aqueuse, ouvre des perspectives pour une compréhension de l’impact des mutations pathogènes et protectrices au niveau moléculaire. / The neurodegenerative Alzheimer's disease (AD) is affecting more than 40 million people worldwide and is linked to the aggregation of the amyloid-β proteins of 40/42 amino acids. Despite many experimental and theoretical studies, the mechanism by which amyloid fibrils form and the 3D structures of the early toxic species in aqueous solution remain to be determined. In this thesis, I studied the structures of the eraly formed oligomers of the amyloid-β peptide and the critical nucleus size of two amyloid-β peptide fragments using either coarse-grained or all-atom simulations. First, at the coarse-grained level, I developed a lattice model for amyloid protein, which allows us to study the nucleus sizes of two experimentally well-characterized peptide fragments (Aβ)16-22 and (Aβ)37-42 of the Alzheimer's peptide (Aβ)1-42. After presenting a comprehensive OPEP force-field parameterization using an on-lattice protein model with Monte Carlo simulations and atomistic simulations, I determined the nucleus sizes of the two fragments. My results show that the nucleation number is 10 chains for (Aβ)16-22 and larger than 20 chains for (Aβ)37-42. This knowledge is important to help design more effective drugs against AD. Second, I investigated the structures of the dimer (Aβ)1-40 using extensive atomistic REMD simulations. This study provides insights into the equilibrium structure of the (Aβ)1-40 dimer in aqueous solution, opening a new avenue for a comprehensive understanding of the impact of pathogenic and protective mutations in early-stage Alzheimer’s disease on a molecular level.

Submitted by PPG Ci?ncia da Computa??o (ppgcc@pucrs.br) on 2018-09-03T14:00:58Z No. of bitstreams: 1 RAFAEL CAUDURO OLIVEIRA MACEDO_DIS.pdf: 6178948 bytes, checksum: 6ed3599e31f122e78b11b322a8c0ac06 (MD5) / Approved for entry into archive by Sheila Dias (sheila.dias@pucrs.br) on 2018-09-04T12:17:04Z (GMT) No. of bitstreams: 1 RAFAEL CAUDURO OLIVEIRA MACEDO_DIS.pdf: 6178948 bytes, checksum: 6ed3599e31f122e78b11b322a8c0ac06 (MD5) / Made available in DSpace on 2018-09-04T12:47:15Z (GMT). No. of bitstreams: 1 RAFAEL CAUDURO OLIVEIRA MACEDO_DIS.pdf: 6178948 bytes, checksum: 6ed3599e31f122e78b11b322a8c0ac06 (MD5) Previous issue date: 2017-08-30 / Prote?nas executam um papel vital em todos os seres vivos, mediando uma s?rie de processos necess?rios para a vida. Apesar de existirem maneiras de determinar a composi??o dessas mol?culas, ainda falta-nos conhecimentos suficiente para determinar de uma maneira r?pida e barata a sua estrutura 3D, que desempenha um papel importante na suas fun??es. Um dos principais m?todos computacionais aplicados ao estudo das prote?nas e o seu processo de enovelamento, o qual determina a sua estrutura, ? Din?mica Molecular. Um aprimoramento deste m?todo, conhecido como Replica Exchange Molecular Dynamics (ou REMD), ? capaz de produzir resultados muito melhores, com o rev?s de significativamente aumentar o seu custo computacional e gerar um volume muito maior de dados. Esta disserta??o apresenta um novo m?todo de otimiza??o deste m?todo, intitulado Filtragem de Dados Anal?ticos, que tem como objetivo otimizar a an?lise p?s-simula??o filtrando as estruturas preditas insatisfat?rias atrav?s do uso de m?tricas de qualidade absolutas. A metodologia proposta tem o potencial de operar em conjunto com outras abordagens de otimiza??o e tamb?m cobrir uma ?rea ainda n?o abordada por elas. Adiante, a ferramenta SnapFi ? apresentada, a qual foi designada especialmente para o prop?sito de filtrar estruturas preditas insatisfat?rias e ainda operar em conjunto com as diferentes abordagens de otimiza??o do m?todo REMD. Um estudo foi ent?o conduzido sobre um conjunto teste de simula??es REMD de predi??o de estruturas de prote?nas afim de elucidar uma s?ries de hip?teses formuladas sobre o impacto das diferentes temperaturas na qualidade final do conjunto de estruturas preditas do processo REMD, a efici?ncia das diferentes m?tricas de qualidade absolutas e uma poss?vel configura??o de filtragem que utiliza essas m?tricas. Foi observado que as temperaturas mais altas do m?todo REMD para predi??o de estruturas de prote?nas podem ser descartadas de forma segura da an?lise posterior ao seu t?rmino e tamb?m que as m?tricas de qualidade absolutas possuem uma alta vari?ncia (em termos de qualidade) entre diferentes simula??es de predi??es de estruturas de prote?nas. Al?m disso, foi observado que diferentes configura??es de filtragem que utilize tais m?tricas carrega consigo esta vari?ncia. / Proteins perform a vital role in all living beings, mediating a series of processes necessary to life. Although we have ways to determine the composition of such molecules, we lack sufficient knowledge regarding the determination of their 3D structure in a cheap and fast manner, which plays an important role in their functions. One of the main computational methods applied to the study of proteins and their folding process, which determine its structure, is Molecular Dynamics. An enhancement of this method, known as Replica-Exchange Molecular Dynamics (or REMD) is capable of producing much better results, at the expense of a significant increase in computational costs and volume of raw data generated. This dissertation presents a novel optimization for this method, titled Analytical Data Filtering, which aims to optimize post-simulation analysis by filtering unsatisfactory predicted structures via the use of different absolute quality metrics. The proposed methodology has the potential of working together with other optimization approaches as well as covering an area still untouched at large by them to the best of the author knowledge. Further on, the SnapFi tool is presented, a tool designed specially for the purpose of filtering unsatisfactory structure predictions and also being able to work with the different optimization approaches of the Replica-Exchange Molecular Dynamics method. A study was then conducted on a test dataset of REMD protein structure prediction simulations aiming to elucidate a series of formulated hypothesis regarding the impact of the different temperatures of the REMD process in the final quality of the predicted structures, the efficiency of the different absolute quality metrics and a possible filtering configuration that take advantage of such metrics. It was observed that high temperatures may be safely discarded from post-simulation analysis of REMD protein structure prediction simulations, that absolute quality metrics posses a high variance of efficiency (regarding quality terms) between different protein structure prediction simulations and that different filtering configurations composed of such quality metrics carry on this inconvenient variance.

No description available.

Les Récepteurs Couplés aux Protéines-G forment la plus importante famille de protéines membranaires chez l’homme et sont impliqués dans de nombreux processus de signalisation cellulaire. Aussi, ils forment un vivier très important de cibles thérapeutiques, déjà identifiées ou potentielles. L’activation d’un RCPG est amorcée par la liaison d’un ligand dans sa partie extra-cellulaire, modifiant ainsi ses propriétés dynamiques intrinsèques. Ces changements structuraux vont alors se répercuter le long des domaines trans-membranaires et promouvoir la dissociation de la Protéine-G hétéro-trimérique, de l’autre côté de la membrane, propageant ainsi le signal au compartiment intra-cellulaire. Ce processus peut être modulé par la liaison de nombreux autres partenaires des RCPGs. Malgré de nombreuses données structurales existantes, ces mécanismes restent encore mal connus à l’échelle moléculaire. Ainsi, la dynamique moléculaire s’est révélée être un outil formidable pour mieux comprendre ces mécanismes. Toutefois, les échelles de taille et de temps requises pour discuter de la dynamique de ces systèmes membranaires limitent ces études aux laboratoires ayant accès à une très grande puissance de calcul. L’objectif des travaux présentés dans ce manuscrit a été de prédire et de mieux comprendre la dynamique d’interaction de différents récepteurs de cette famille avec leurs partenaires, en développant un protocole de dynamique moléculaire, peu coûteux en ressources de calcul, combinant le champ de forces gros-grains MARTINI à un protocole de dynamique moléculaire « Replica-Exchange ».Dans un premier temps, nous présentons la validation de notre protocole pour la prédiction de la liaison de peptides à leur récepteur avec l’étude des peptides Neurotensine, agoniste du Récepteur de la Neurotensine-1, et CVX15, antagoniste du Récepteur Chemokine C-X-C de type-4. Nous montrons également que notre protocole est capable de prédire la sélectivité de plusieurs peptides dérivés de la Neurotensine envers plusieurs récepteurs sauvages et mutés, ne présentant qu’un résidu de différence.Dans un second temps, nous nous sommes intéressés à la dynamique de formation d’un hétéro-dimère de RCPGs impliquant le Récepteur de la Ghréline et le récepteur de la Dopamine D2, couplés aux protéines Gq et Gi respectivement. Ce modèle validé au laboratoire par des mesures LRET montre une interface impliquant une forte complémentarité entre les protéines-G. En se basant sur notre modèle, nous avons conçu et synthétisé des peptides inhibiteurs de la formation de cet hétéro-dimère de protéines-G.Enfin, nous présentons d’autres exemples d’applications de notre protocole et comment il peut être utilisé de concert avec l’expérience avec : la prédiction de la liaison de toxines de serpents aux Récepteurs de la Vasopressine-1a et V2 ; la prédiction de la liaison des peptides Ghréline et Leap2 au Récepteur GHSR-1a et la prédiction de la sélectivité de couplage de différents récepteurs aux peptides C-terminaux de la sous-unité α des protéines-G. / G-Protein Coupled Receptors form the largest family of human membrane proteins and are involved in many cellular signaling processes. Thus, they constitute a pool of already identified or potential pharmacological targets. The activation of a GPCR starts with the binding of a ligand in its extra-cellular part, further modifying its intrinsic dynamical properties. These structural rearrangements are then transmitted along the transmembrane domains and promote the dissociation of the G-protein on the other side of the bilayer, thus propagating the signal into the intra-cellular compartment. This activation process can be modulated by the binding of many other partners of GPCRs. Despite many structural data now available, these mechanisms are still badly known at the molecular scale. In agreement, molecular dynamics simulations appear to be a method of choice to get a better description of these mechanisms. Nevertheless, the size and the time scales required for the simulation of these membrane systems limit such studies to laboratories having access to large computational facilities.The objective of this work was to predict and get a dynamical view of the interactions of several GPCRs with their partners, by developing an affordable molecular dynamics protocol that combines the coarse-grained MARTINI force field to Replica-Exchange MD simulations.In a first step, we validated our protocol by showing its ability to predict the dynamical binding of peptides to their receptors, through the study of Neurotensin, an agonist of the Neurotensin-1 receptor and CVX15, an antagonist of the CXCR4 chemokine receptor. We also show that the same protocol is able to predict the selectivity of several Neurotensin derived peptides against several wild-type/mutated receptors differing by a single residue.In a second step, we were concerned by the dynamical assembly of a GPCR heterodimer involving the Ghrelin and the Dopamine D2 receptors, respectively coupled to Gq and Gi proteins. Our model was validated by LRET measurements confirming a large protein:protein interface and a high complementarity between G-proteins. Based on this model, we designed and synthesized some peptides able to inhibit the assembly of this G-proteins heterodimer.Finally, we describe other applications of our protocol and how it can be employed and confronted to experiments to : predict the dynamical binding of toxins from snake’s venom to the Vasopressin-1a and Vasopressin-2 receptors ; predict the binding of the Ghrelin and Leap2 peptides to their GHSR-1a receptor and predict the coupling selectivity of several receptors to peptides mimicking the C-terminus of the α subunit of G-proteins.