Program

Preliminary Workshop Program

Tuesday 06.06.2006

08:00 Pick up in Aachen

09:00 Registration

10:00 - 10:15 "Supercomputing at NIC", Thomas Lippert, Director NIC

10:15 - 10:30 "From Computational Biophysics to Systems Biology", Ulrich Hansmann

Morning Session (Chair: U.H.E Hansmann)

10:30 - 11:30 Harold Scheraga, The Two Aspects of the Protein Folding Problem

11:30 - 11:40 Organizational remarks

11:40 - 12:00 Coffee break

12:00 - 12:30 Adam Liwo, Optimization of the united-residue UNRES force field for Langevin dynamics

12:30 - 14:00 Lunch

Afternoon Session (Chair: J.H. Meinke)

14:00 - 14:45 Tetsu Narumi, A High-Speed Special-Purpose Computer for Molecular Dynamics Simulations: MDGRAPE-3

14:45 - 15:15 Cezary Czaplewski, Multiplexed-replica exchange molecular dynamics with UNRES force-field as an effective method for exploring the conformational energy landscape of proteins

15:15 - 15:45 Coffee break

15:45 - 16:15 Mariusz Makowski, Simple physics-based analytical formulas for the potentials of mean force of the interaction of amino-acid side chains in water. Tests with simple spherical systems

16:15 - 16:45 Stanislaw Oldziej, Optimizing the UNRES force field for the prediction of proteins

16:45 - 17:15 Michael Bachmann, Reduction to the simplest - The complexity of minimalistic heteropolymer models

17:15 - 17:45 Dominik Gront High accuracy method to protein backbone reconstruction

18:00 - 20:00 Welcome Reception

20:00 Bus to Aachen

Wednesday 07.06.2006

08:00 Pick up in Aachen

Morning Session (Chair: T. Narumi)

09:00 - 9:45 Wolfgang Wenzel, De novo protein structure prediction and folding

09:45 - 10:15 Martin Mönnigmann, Protein Loop Structure Prediction With Flexible Stem Geometries

10:15 - 10:45 Walter Nadler, Molecular Transport Through Channels and Pores: Effects of In-Channel Interactions and Blocking

10:45 - 11:15 Coffee break

11:15 - 12:00 Gordon Crippen Cluster Distance Geometry and Protein Folding

12:00 - 12:30 Giovanni La Penna, Modelling protein-DNA interactions in chromatin

12:30 - 14:00 Lunch

Afternoon Session (Chair: Adam Liwo)

14:00 - 14:45 Anders Irbäck, Protein folding and aggregation studied using an all-atom model with a simplified energy function

14:45 - 15:15 Inta Liepina, Molecular dynamics study of two ABL-SH3 domain peptides amyloid formation

15:15 - 15:45 Coffee break

15:45 - 16:15 Siegfried Höfinger A Very General Solvation Model for BioMolecular Simulation

16:15 - 16:45 Magdalena Gruziel, From microscopic Monte-Carlo simulations to macroscopic solvation models

17:00 - 19:00 Poster Session (beverages and snacks provided)

19:00 Bus to Aachen

Thursday 08.06.2006

08:00 Pick up in Aachen

Morning Session (Chair: A. Irbäck)

09:00 - 09:30 Matthias Stein, Integrating Structural and Kinetic Enzymatic Information in Systems Biology

09:30 - 10:00 Jesus Izaguirre, Inferring Protein-Protein Interactions from Data and Domain Docking

10:00 - 10:45 Roland Eils, Computational Systems Biology

10:45 - 11:15 Coffee break

11:15 - 11:45 Anton Feenstra, Sub-type Specific Sites for SMAD Receptor Binding Identified by Sequence Comparison using "Sequence Harmony"

11:45 - 12:15 Daniele Dell'Orco, In silico screening of mutational effects on Enzyme-Inhibitor affinity: a docking-based approach

12:15 - 12:45 Francesco Guerrieri, Ab initio simulations of the Cu binding sites on the N-terminal

12:45 - 14:00 Lunch

Afternoon Session (Chair: O. Zimmermann)

14:00 - 14:45 Ruhong Zhou, Dewetting Transition and Hydrophobic Collapse in Protein Aggregates

15:00 Leaving for Aachen
15:45 Tour of the cathedral
17:30 Tour of downtown Aachen
19:00 Dinner in Aachen

Friday

08:00 Pick up in Aachen

Morning Session (Chair: G. Crippen)

09:00 -9:45 Yuko Okamoto, Generalized-Ensemble Algorithms and Membrane Protein Structure Predictions

09:45 - 10:15 Yuguang Mu, A new method of replica exchange method for protein folding in explicit water

10:15 - 10:45 Caroline Taylor, Using simple fluid wetting as a model for cell spreading

10:45 - 11:15 Coffee break

11:15 - 12:00 A. Kolinski, Modeling protein structure, dynamics and thermodynamics with reduced representation of conformational space

12:00 - 12:30 Wolfhard Janke, Adsorption phenomena at organic-inorganic interfaces

12:30 - 14:00 Lunch

Afternoon Session (Chair: U.H.E. Hansmann)

14:00 - 14:30 Phanourios Tamamis, Conformational analysis of compstatin with molecular dynamics simulations in explicit water

14:30 - 15:15 Avijit Ghosh, From Simulation to Therapy: A Systems Biology Approach to Oncogene Detection and Drug Design

15:15 - 15:45 Coffee break

15:45 - 16:30 Tour of NIC Supercomputers

17:00 - 20:30 Social: Opening game of the World Cup 2006 Germany against Costa Rica (Begin 18:00 ZDF)

20:30 Bus to Aachen leaves after the game (Return to Aachen in time to watch Poland vs. Ecuador)

Abstracts

The Two Aspects of the Protein Folding Problem, H.A. Scheraga, A. Liwo, S. Ołdziej, C. Czaplewski, M. Khalili

Using a physics-based approach, based on an empirical potential energy function, it is now possible to compute not only the native structure but also the structural pathways from the unfolded to the folded (native) structure of a globular protein. These are the two aspects of the protein folding problem. The folding algorithm involves two basic ingredients: (i) development of a suitable potential energy function and (ii) global optimization of this function by an efficient search of the conformational energy surface of the protein. The computational methodology evolved from an all-atom description of the polypeptide chain, together with minimization algorithms, to a simpler united-residue (UNRES) model which was initially optimized with a Conformational Space Annealing algorithm (CSA) that was subsequently replaced by a Langevin molecular dynamics treatment of the UNRES model. The evolution of this model, its performance in successive CASP exercises, and its ultimate use to compute kinetic pathways and native structures will be described.

Modeling protein structure, dynamics and thermodynamics with reduced representation of conformational space, Andrzej Kolinski

We review recent applications of our protein modeling methodology based on reduced representation of protein conformational space. The reduced representation of the CABS model employs up to four interaction centers per residue (alpha carbon, beta carbon, the center of mass of the side group and the center of peptide bond). Carefully designed, tuned and tested force field of CABS consists of several potentials of mean-force derived from statistical analysis of structural regularities seen in known protein structures. The sampling of conformational space of the model proteins bases on various variants of the Monte Carlo method, including very efficient muticopy (Replica Exchange Monte Carlo) algorithms. A number of supplementary bioinformatics tools have been developed to handle data processing and analysis of the large scale simulations of protein systems via CABS.

CABS methodology proven to be one of the best methods for protein structure prediction, from comparative modeling to de novo folding. It has been clearly demonstrated during the sixth CASP (Critical Assessment of protein Structure Prediction) community-wide experiment. The groups employing the CABS-based methodology scored among the best.

The newest applications of the CABS modeling technology include: study of protein folding thermodynamics and pathways, structure prediction based on sparse and inaccurate experimental data and prediction of protein-protein interactions or flexible ligand docking. The CABS reduced model could be easily integrated with the old-atom approaches providing solid starting point for fast multiscale simulations of large biomolecular systems.

Protein folding and aggregation studied using an all-atom model with a simplified energy function,Anders Irbäck

We present results from all-atom Monte Carlo simulations of protein folding and aggregation, which were performedusing a novel simplified interaction potential. This model has been used to study (i) the folding of several alpha-helical and beta-sheet peptides with about 20 amino acids, (ii) the oligomerization of a 7-amino acid fragment of the Alzheimer's Abeta peptide, and (iii) the mechanical and thermal unfolding of the 76-amino acid protein ubiquitin. The different systems were studied without changing any parameter of the model.

De novo protein structure prediction and folding, A. Verma, S. Murthy, K. H. Lee, W. Wenzel

Protein structure prediction and the elucidation of the protein folding process remain important challenges for biophysical chemistry. According to Anfinsons thermodynamic hypothesis many proteins are in thermodynamic equilibrium with their environment, their unique native conformation is global optimum of the free-energy surface. Using efficient stochastic optimization methods we are able to determine the global optimum of the complex protein free-energy surface orders of magnitude faster than by traditional simulation techniques.

We have parameterized an all-atom free-energy forcefield for proteins (PFF01/02), which is based on the fundamental biophysical interactions that govern the folding process. We have also developed, or specifically adapted, efficient stochastic optimization methods to simulate the protein folding process.

With this approach we were able to predictively and reproducibly fold the trp-cage protein (23 amino acids) , the villin headpiece (36 amino acids), the HIV accessory protein (40 amino acids) and beta-hairpin proteins (14-20 amino acids) in simulations starting from random initial conformations.

We have also investigated methods to extend our approach to larger proteins by combining our free energy model with heuristic techniques that generate large libraries of protein conformations on the basis of the amino acid sequence. When we ranked such decoy sets for 30 different proteins according to their energy in our model, we find that near-native conformations are selected for all high-quality decoy-sets. For low-quality decoy sets, the approach generates usable low-resolution models in over 80% of the cases, but still has difficulty treating disulfide-bridged proteins, protein-protein complexes and proteins which are stabilized only in complex with other molecules.

A High-Speed Special-Purpose Computer for Molecular Dynamics Simulations: MDGRAPE-3, T. Narumi, Y. Ohno, N. Futatsugi, N. Okimoto, A. Suenaga, R. Yanai, and M. Taiji

We are developing a high-speed computer, MDGRAPE-3, dedicated for molecular dynamics simulations. It will be completed during this year and be the first Peta Flops machine. The MDGRAPE-3 is a special-purpose computer for force calculation between particles. Coulomb and van der Waals force calculations are accelerated by the MDGRAPE-3, and the other calculations are performed on the usual general-purpose computer. With this architecture, we can get higher performance, better performance by cost, and lower power by performance than those of any other computers. In the talk, hardware and software of the MDGRAPE-3 will be presented as well as recent simulation results, such as folding simulation of a small protein, simulations of RNA polymerase, and free energy calculation between protein and other molecules.

Dewetting Transition and Hydrophobic Collapse in Protein Aggregates, R. Zhou

We have performed molecular dynamics simulations of the collapse of a two-domain protein BphC enzyme, and the melittin tetramer, to examine how water molecules mediate hydrophobic collapse in proteins. For the two domain collapse, water depletion and hydrophobic collapse occur on a nanosecond time scale, two orders of magnitude slower than that found in the collapse of idealized paraffin-like plates. When the electrostatic protein-water forces are turned off, a dewetting (water drying) transition occurs in the inter-domain region and the collapse speeds up by more than an order of magnitude. When attractive van der Waals forces are further turned off, the dewetting in the inter-domain region becomes more profound, and the collapse speed mimics that of the idealized plates. In the melittin tetramer folding, we observed a surprising water drying transition inside a nanoscale channel formed by the tetramer, with a channel size of up to 2-3 water diameters. This transition, although occurring on a microscopic length scale, is reminiscent of the first order phase transition from liquid to vapor. We have found that this drying transition is very sensitive to single mutations and such mutations in the right locations can switch the channel from being dry to being wet. Therefore, even in the presence of the polar protein backbone, sufficiently hydrophobic protein surfaces can induce a liquid-vapor transition which can then provide an enormous driving force towards further collapse.

In this talk, I will also give a very brief overview of the IBM BlueGene project, including its hardware design and protein science program. Some preliminary results on protein lysozyme misfolding/aggregation as well as some novel sampling methods will also be presented.

Generalized-Ensemble Algorithms and Membrane Protein Structure Predictions, Yuko Okamoto

Generalized-ensemble algorithm is a generic term for a simulation method that is based on non-Boltzmann weight factors and performs a random walk in energy space so that the multiple-minima problem can be overcome. Recently, we have applied one of the generalized-ensemble algorithms, namely, the replica-exchange method (or parallel tempering) to the prediction of membrane protein structures. Our method consists of two parts. In the first part, amino-acid sequences of the transmembrane helix regions are obtained from one of existing WWW servers such as SOSUI. In the second part, we perform a replica-exchange simulation of these transmembrane helices with some constraints and identify the predicted structure as the global-minimum-energy state. We have tested the method with the dimeric transmembrane domain of glycophorin A and bacteriorhodopsin. In this talk, I will first explain generalized-ensemble algorithms and then present the results of membrane protein structure predictions.

From Simulation to Therapy: A Systems Biology Approach to Oncogene Detection and Drug Design, D. Pant, A. Kumar, R. Zou, A. Ghosh

In silico models of signal transduction pathways have been highly successful in describing, quantitatively, how complex protein networks govern overall cell function. Moreover, such models have been successful not only in elucidating function, but moreover malfunction. We have recently examined the nature of oncogenic transformation of normal cell lines with a focus on the MAPK signal transduction pathway. Activation of the Extracellular signal Regulated Kinases (ERK1/2; p42/p44 MAPK) is one of the most extensively studied signaling pathways in part because it occurs downstream of oncogenic RAS. By using ERK phosphorylation as an "output signal", the method analyzes experimentally determined kinetic data and predicts putative oncogenes and tumor suppressor gene products impacting the RAS/ MAPK module. This analysis has identified several modifiers of ERK/ MAPK activation described previously in the literature. In addition, several novel enzymes are identifie d which have not previously described to affect ERK/ MAPK phosphorylation. Importantly, the nonlinear analysis enables a ranking of modifiers of MAPK activation predicting their relative importance in RAS-depThe analysis described has lead to the development of a set of virtual transformed cells that have been the target of a complementary drug design algorithm recently developed by our group. The algorithm generates a quantitative ranking of target enzymes which inhibit the transformation process. The effect of adding an inhibitor to each protein within the signal cascade is measured and ranked. The inhibitor, a virtual drug, is constructed by specifying its parameters: concentration and binding affinity. Results show that a Ras inhibitor is the most effective inhibitor, ranking at the lowest concentrations and highest binding affinities in relation to the other effective drugs designed for the oncogenic pathways. Furthermore, we suggest other targets that are expected to be effective and suitable for further clinical study are: Gef, Raf, Pkc, and Mek. Results also show that certain classes of calcium blockers may additionally be effective in controlling proliferation.

Cluster Distance Geometry and Protein Folding, G. Crippen

Cluster distance geometry is a recent generalization of distance geometry whereby protein structures can be described at even lower levels of detail than one point per residue. Protein conformations can be summarized in terms of alternative contact patterns between clusters, where each cluster contains four sequentially adjacent residues. A very simple potential function involving 210 adjustable parameters can be determined that favors the native contacts of 31 small monomeric proteins over their respective sets of nonnative contacts. This potential then favors the native contacts for 698 small proteins, even though they have low sequence identity with the training set.

Computational Systems Biology, Roland Eils

Computational systems biology [1] is a field that aims at a system-level understanding by analyzing biological data using computational techniques. The explosive progress of genome sequencing projects and the massive amounts of data generated by high-throughput experiments in DNA microarrays, proteomics, and metabolomics advances this field in a bidirectional, but dependent fashion. As the need for a complete quantitative part list in biology is recognized, the understanding develops that living system cannot be understood by studying just individual parts. Under the guiding vision of systems biology, sophisticated computational methods are currently developed to analyze the data generated by this new technology in a systematic fashion, unraveling complex and networked biological phenomena, and modeling processes that take place in cells, tissues and organisms. The construction and testing of quantitative representations and models will be possible through the collaborative input of experimental and theoretical biologists working together with system analysts, computer scientists, mathematicians, engineers, physicists, as well as physicians to contend creatively with the hierarchical and nonlinear nature of cellular systems, while bioengineers will maintain a focus on directing the research results towards developing and improving cell-based, biotechnological processes. In my presentation I will give a general overview on the field of systems biology. Further, I will report on recent work on modeling of large signal transduction networks in mammalian cells [2].

[1] Kriete A and Eils R. (2005) Computational Systems Biology, Elsevier. [2] Bentele M, Lavrik I, Ulrich M, Stosser S, Heermann DW, Kalthoff H, Krammer PH, Eils R. (2004) Mathematical modeling reveals threshold mechanism in CD95-induced apoptosis . J Cell Biol. 166(6):839-51.