Skip to Main Content
From Computational Biophysics to Systems Biology (CBSB13)
May 19-21, 2013  Norman, Oklahoma
OU link

Invited Speakers:

Keynote Speakers:

Invited Speakers:

    Hans Frauenfelder
    Los Alamos National Laboratory

    Motions are crucial for the function of proteins, for instance for the opening and closing of channels [1]. Three types of fluctuations have been found so far that are important, α-fluctuations that originate in the bulk solvent, β-fluctuations that come mainly from the hydration shell that surrounds the protein, and thermal vibrations that come from the entire system. Studies of these fluctuations with the Mössbauer effect, neutron scattering, and dielectric relaxation spectroscopy provide new insights [2]. They challenge the accepted wisdom, provide a new picture of how these techniques work in complex systems, and give a view of the low tiers of the energy landscape.

    [1] H. Frauenfelder et al., Proc. Natl. Acad. Sci. USA 106, 5129 (2009).
    [2] R. D. Young et al., Phys. Rev. Lett. 107, 158102 (2011).
    Back to the top of the page

  • Experimental-Computational Discoveries Achieved Through the Blue Waters and Titan Petascale Computers
    Klaus Schulten
    University of Illinois Urbana-Champaign

    Summary: Earlier this year, 2013, two remarkable computers have become available to computational scientists in the US, National Science Foundation-funded Blue Waters and Department of Energy-funded Titan. The machines, offering nearly 300,000 cores and GPU acceleration, permit molecular dynamics and molecular analysis calculations with NAMD and VMD of unprecedented size and time scales. BlueWaters and Titan provide, in particular, new opportunities to link experiment and theory in molecular cell biology and, in fact, already during the test period remarkable computational biology discoveries resulted from the two machines. This lecture reports on three of these discoveries.
    1. Permitting extensive experimentation and sampling, molecular dynamics simulations demonstrated that synaptotagmin, a key agent in Ca++-triggered neurotransmitter release, should act in membrane sculpting in a Ca++ activated conformation, which differs from the crystallographically observed structure.
    2. Permitting extensive transition path sampling molecular dynamics simulations investigated the mechanism of a homo-hexameric molecular motor, hexameric helicase, that translates along single-stranded RNA. The simulations revealed that the sequence of ATP binding, ATP hydrolysis, ADP + Pi release reactions affect only insignificantly protein conformations, but mainly induce a rearrangement of the six subunits relative to each other through alteration in surface-surface interactions. The resulting motion of the hexamer of subunits alters the interaction and positioning of lysine groups interacting with the central RNA such that the RNA is moved by one base per ATP hydrolysis relative to the hexamer.
    3. Molecular dynamics flexible fitting (MDFF) combining NMR, crystallography, and EM data with molecular modeling solved the atomic level structure of an entire HIV virus capsid. The protein data base (pdb) entry of the capsid with 3 million atoms is the largest pdb entry yet and documents the malleability of the single type capsid protein that assembles into a rather heterogeneous system of about 1300 proteins realizing a non-symmetric distribution of local curvature.
    Back to the top of the page

  • The iPlant Collaborative: A Life Sciences Cyberinfrastructure for the 21st Century
    Dan Stanzione
    University Texas at Austin

    The iPlant Collaborative (iPlant) is a large scale United States National Science Foundation (NSF) funded project that aims to create an innovative, comprehensive, and foundational cyberinfrastructure in support of plant biology research (and now, animal science as well). iPlant supports the diverse fields and workflows that comprise plant biology by building an open, flexible, extensible platform. iPlant provides services ranging from large scale storage and computation, to application programmer interfaces, to web-based environments for end users. In this talk, an overview of iPlant's CI capabilities will be presented, along with a discussion of how these resources take advantage of the newly-deployed cutting edge supercomputing systems at the Texas Advanced Computing Center, followed by a brief discussion of changes in computing architecture and the implications for bioinformatics applications.
    Back to the top of the page

  • Lipids and hydrogen bonding in membrane protein function
    Ana Nicoleta Bondar
    Freie Universität Berlin

    Hydrogen-bond dynamics and changes in the protonation state couple to the conformational dynamics of membrane proteins. Based on extensive investigations of the dynamics of several membrane proteins, we propose that inter-helical hydrogen bonding and hydrogen bonding to lipids ensures efficient means of long-distance conformational coupling. In the absence of perturbations (mutation, protonation change), inter-helical hydrogen bonds help stabilize protein conformation. Once the protein is perturbed, however, dynamical hydrogen bonds can rearrange rapidly and help stabilize a new protein conformation. Hydrogen bonding to lipids not only affects the local structure and dynamics of the protein, but can influence the dynamics of inter-helical hydrogen bonds, effectively coupling membrane protein function to the membrane environment.
    Back to the top of the page

  • Macromolecular solution properties by analytical ultracentrifugation
    Borries Demeler
    University of Texas Health Science Center at San Antonio

    Analytical ultracentrifugation (AUC) is a powerful technique used to characterize the solution behavior of macromolecules. Being able to work in the solution state offers many advantages, because many conditions of a reaction can be easily modulated, such as pH, ionic strength, temperature, oxidation state, the presence of small molecules, and other binding partners. This allows the researcher to investigate DYNAMIC processes, study the effects of mutations, investigate oligomerization states, and follow complex multi-domain protein assembly. Unlike other techniques, AUC has virtually no size restrictions, and is suitable for a wide variety of samples and sample conditions. Advanced optical detectors and new computational methods allow us to obtain high resolution information describing sample composition (size, anisotropy, and partial concentration) as well as to measure binding strength of interactions between one or more binding partners. AUC is based on first principles and thus does not require any standards. In this talk I will present an overview of the technique, discuss some of the analysis methodology utilizing high performance computing, and review applications in AUC using illustrative examples.
    Back to the top of the page

  • A Novel Optimization Framework for the Design of Proteins with Post-Translational Modifications and Unnatural Amino Acids
    George A. Khoury, James Smadbeck, and Christodoulos A. Floudas
    Department of Chemical and Biological Engineering, Princeton University

    Most proteins being used for drug applications contain post-translational modifications (PTMs) [1]. It can become prohibitively expensive and time-consuming to experimentally modify and test the effects of large numbers of modifications. To enable academic and industrial researchers to minimize time and monetary cost searching for new functional designs, in this work we present ModDesigner, a method to in silico design proteins and peptides with post-translational modifications and unnatural amino acids.

    We develop a new integer linear optimization design formulation to introduce the modified amino acids, drawing on our recently developed forcefields [2]. Given a ligand or an ensemble of ligand conformations of a peptide bound in a receptor protein, the formulation maximizes the favorable (Van der Waals and hydrogen bond) contacts and minimizes steric clashes while maintaining or enhancing the binding free energy. To reduce the combinatorial complexity of the problem, logical restraints to preserve the residue-wise charge and hydrophobicity are applied to reduce the number of allowed modifications in each position. For each configuration, the initial contacts, clashes, and hydrogen bonds are populated. Next, for each configuration, each design position is modified, a local energy minimization is performed, and the new contacts, clashes, hydrogen bonds and interaction energy are populated. Finally, the ILP is solved to global optimality, generating a rank-ordered list of designs using integer cuts. The importance of using a flexible template in using these design metrics is demonstrated using an ensemble of conformers generated from a molecular dynamics simulation.

    The algorithm is designed to serve as Stage 3: Modification Selection of our generalized de novo design framework [3,4], with Stage 1 being an optimization driven Sequence Selection, and Stage 2 being Fold Specificity and statistical mechanics-based Approximate Binding Affinity calculations. We first benchmark the forcefield which is based on only physics ability to predict IC50 values. Next, we describe our efforts to use the method to design new variants of Compstatin to inhibit Complement C3c beginning with the extremely potent Compstatin analog [5], Variant E1. In all efforts we begin with template sequences and structures containing only the 20 amino acids discovered to experimentally bind and inhibit their target receptors [6-8]. In this way, the modified amino acids serve to fine-tune specificity and potentially enhance the affinity of a given inhibitor.

    1. Walsh C. Posttranslational modification of proteins: expanding nature's inventory. Englewood, Colo.: Roberts and Co. Publishers; 2006. xxi, 490 p. p.
    2. Khoury GA, Thompson J, Smadbeck J, Floudas CA. ForcefieldPTM: Development and Testing of a First Generation AMBER Forcefield for Post-Translational Modifications In Preparation.
    3. Bellows ML, Fung HK, Floudas CA. Recent Advances in De Novo Protein Design. Process Systems Engineering: Wiley-VCH Verlag GmbH & Co. KGaA; 2011. p 207- 232.
    4. Smadbeck J, Bellows-Peterson ML, Khoury GA, Taylor MS, Floudas CA. Protein WISDOM: A Workbench for In Silico De novo design of bioMolecules. Journal of Visualized Experiments In Press.
    5. Mallik B, Katragadda M, Spruce LA, Carafides C, Tsokos CG, Morikis D, Lambris JD. Design and NMR Characterization of Active Analogues of Compstatin Containing Non- Natural Amino Acids. Journal of Medicinal Chemistry 2004;48(1):274-286.
    6. Bellows ML, Taylor MS, Cole PA, Shen L, Siliciano RF, Fung HK, Floudas CA. Discovery of Entry Inhibitors for HIV-1 via a New De Novo Protein Design Framework. Biophysical Journal 2010;99(10):3445-3453.
    7. Bellows ML, Fung HK, Taylor MS, Floudas CA, López de Victoria A, Morikis D. New Compstatin Variants through Two De Novo Protein Design Frameworks. Biophysical Journal 2010;98(10):2337-2346.
    8. López de Victoria A, Gorham RD, Bellows-Peterson ML, Ling J, Lo DD, Floudas CA, Morikis D. A New Generation of Potent Complement Inhibitors of the Compstatin Family. Chem Biol Drug Des 2011;77(6):431-440.
    Back to the top of the page

  • Probing functional dynamics in enzymes using simulations
    Donald Hamelberg
    Georgia State University

    Molecular dynamics is now commonly used as a complementary tool to experiments in understanding the dynamical behavior of biomolecules. Application of advanced sampling methods, such as accelerated molecular dynamics, to probe long time scale dynamics is fast becoming indispensible in the field, since conformational dynamics is generally accepted to be important in enzyme catalysis. However, the precise role of conformational dynamics in enzymes remains unresolved. We present a rationale of how enzyme dynamics is coupled to the reaction step and affects the catalytic rate. Our atomistic simulations provide insights into the general interplay between enzyme conformational dynamics and catalysis from an atomistic perspective, and the results are in notable agreement with experiments. We further discuss our efforts in finding a fast, yet accurate, advance sampling method to probe long time scale dynamics in biomolecules.
    Back to the top of the page

  • Correlation networks of rhinovirus capsid dynamics
    Carol Post
    Purdue University

    Human rhinovirus (HRV) and other members of the enterovirus genus bind small-molecule antiviral compounds in a cavity buried within the viral capsid protein VP1. These compounds block the release of the viral protein VP4 and RNA from inside the capsid during the uncoating process. In addition, the antiviral compounds prevent “breathing” motions, the transient externalization of the N-terminal regions of VP1 and VP4 from the inside of intact viral capsid. The site for externalization of VP1/VP4 or release of RNA is likely between protomers, distant to the binding cavity for antiviral compounds. Molecular dynamics (MD) simulations were conducted to explore how the antiviral compound, WIN 52084, alters properties of the HRV 14 capsid through long-distance effect. The Pearson correlation coefficient of atomic positions, typically used to search MD trajectories for correlated motions, failed to detect any long-range correlations. Nonetheless, an alternative metric of the radial distance uncovered a number of long-range correlations of capsid dynamics, which have not been previously recognized. In the absence of WIN, correlated radial motion is observed between residues separated by as much as 85 Angstroem, a remarkably long distance. We developed a framework to analyze these long-distance, concerted motions using a network based on the radial correlation coefficients. The most frequently populated path segments of the network were localized near the fivefold symmetry axis and included those connecting the N termini of VP1 and VP4 with other regions, in particular near twofold symmetry axes, of the capsid. Moreover, the presence of WIN destroys this radial correlation network, suggesting that the underlying motions contribute to a mechanistic basis for the initial steps of VP1 and VP4 externalization and uncoating.
    Back to the top of the page

  • Biological Applications of the UNRES Force Field
    Harold Scheraga
    Cornell University

    Our UNited RESidue force field (UNRES), developed with Adam Liwo and coworkers, has been made available on a website ((http://UNRES.PL). It has undergone blind tests in the recent CASP10 exercise, with interesting success so that it is now being applied in studies of biological complexes. In addition, the coarse-grained UNRES philosophy has been extended as NARES-2P to treat DNA and RNA (Y. He, M. Maciejczyk, S. Oldziej, H.A. Scheraga, and A. Liwo, Phys. Rev. Letters, in press). These recent results will be discussed in detail in this CBSB13 workshop talk.
    Back to the top of the page

  • Generality of parallel replica dynamics and applicability to biological systems
    Arthur Voter
    Los Alamos National Laboratory

    Many important processes in chemistry, physics, materials science, and biology take place on time scales that exceed what can be accessed directly with molecular dynamics simulation. Often the dynamics on these longer time scales consists of infrequent transition events that take the system from one state to another. Our research program has been aimed at developing methods for extending the accessible molecular dynamics simulation time for this type of system, methods that make as few approximations as possible. Our focus is on an accelerated molecular dynamics (AMD) approach, in which we let the trajectory itself find an appropriate way to escape from each state, but we coax it into doing so more quickly. One of these AMD methods, parallel replica dynamics (ParRep), achieves this by running many replicas of the system in parallel. Implemented carefully, ParRep can give arbitrarily accurate dynamics for infrequent event systems, and can give substantial computational speedup, up to the number of processors, when the events are very infrequent. A requirement for ParRep is that the distribution of first-passage times is exponential, as it is naturally for deep states that lose their memory long before the next escape event. A recent mathematical analysis in the context of overdamped Langevin dynamics [ C. Le Bris, T. Lelievre, M. Luskin, and D. Perez, Monte Carlo Methods and Applications 18, 119 (2012)] has shown that the ParRep method is even more general than we previously realized. In essence, a system that does not have exponentially distributed first-passage times can be converted to one that does, simply by applying the ParRep dephasing procedure for a sufficiently long time, yielding the so-called quasi-stationary distribution (QSD). This opens the possibility of using ParRep on a much broader range of complex systems than we previously thought possible, such as systems where each \\}"state\\\" actually consists of many (perhaps a huge number of) substates, and for which there may not be a good separation of time scales between the equilibration time within the state and the time for escape from the state. I will describe the ParRep method, explain how it is applied in this new QSD-based context, and discuss the implications for treating biological systems, which are the subject of this conference.
    Back to the top of the page

  • Wanted: The Best Models in Systems Biology
    Eberhard Voit
    Georgia Institute of Technology

    One of the grand challenges of computational systems biology is the translation of an actual biomedical phenomenon into a computational format that can be used for interrogation, analysis, manipulation, and optimization. This translation not only requires an understanding of the functional structure and regulation of the actual system underlying the phenomenon, but also the choice of effective mathematical formulations. Alas, nature has not provided us with a list of optimal functions or a handbook of best practices, with the consequence that we simply do not know how to select optimal mathematical or computational representations. In the first part of this presentation, I will identify the challenges faced during model selection and present power-law models as reasonable defaults, at least to get started with model design and analysis. In the second part I will demonstrate to what degree we may obtain a glimpse of the actual shapes of processes within the context of biochemical and metabolic pathway systems.
    Back to the top of the page

  • Microsecond simulations at electronic structure quality: heading toward a biological force field from accurate water models
    Feng Wang
    University of Arkansas

    Developing accurate force fields that are efficient for biological free energy simulations is a grand challenge. High force field accuracy requires sophisticated energy expressions to model many-body effects that are consequence of quantum mechanics. Without relying on complicated many-body energy expressions, the adaptive force matching (AFM) method developed in the Wang group robustly models the quantum mechanical potential energy surface by systematically optimizing the energy expressions for each specific system. Acting as a bridge that connects quantum-mechanical potential energy surface to simulations at extended time and length scales, AFM force fields can make reliable predictions of key experimental properties. In this talk, we present studies of free energy landscape of the ubiquitous biological solvent, water. Microsecond simulations are reported investigating ice-liquid and liquid-liquid phase transitions in water.
    Back to the top of the page

  • Continued Sampling: Using data aware feedback and control for peptide conformational search
    Tom Woolf
    John Hopkins University

    Sampling till completion is a new framework for conformational search that uses a coupled feedback between running computational dynamics and a large database. By using this feedback we can get improved convergence as well as more efficient sampling. We have applied this concept to sampling of peptide degrees of freedom and the first part of the presentation will highlight the challenges in making this idea work by addressing both hardware and software issues. This includes efforts to create rapid query evaluations and our use of coupled CPU and GPU systems. The second part of the presentation will focus on the nature of the efficiency gains and how the concept may be helpful for sampling within the Onsager-Machlup framework for understanding conformational transitions. This enables us to view certain degrees of freedom as more important for the transitions than others and ties into dynamic importance sampling and effective transfer entropy approaches.
    Back to the top of the page