[Basel Computational Biology Conference 2004]

  Abstracts
 
 

 

Keynote Lecture: Computational Physiology and the Physiome Project

Peter Hunter (Bioengineering Institute, University of Auckland, New Zealand)

The IUPS Physiome Project (I) is an internationally collaborative open-source project to provide a public domain framework for computational physiology, including the development of modeling standards, computational tools and web-accessible databases of models of structure and function at all spatial scales [1,2,3]. It aims to develop an infrastructure for linking models of biological structure and function in human and other eukaryotic physiology across multiple levels of spatial organization and multiple time scales. The levels of biological organisation, from genes to the whole organism, includes gene regulatory networks, protein-protein and protein-ligand interactions, protein pathways, integrative cell function, tissue and whole organ structure-function relations, and finally the integrative function of the whole organism.

The project requires the creation of web-accessible databases of mathematical models of structure and function at spatial scales which encompass nano-scale molecular events to metre-scale intact organ systems, a range of 10 9 , and temporal scales from Brownian motion (microseconds) to a human lifetime (10 9 s), a range of 10 15 . Clearly this cannot be represented by one model but rather a hierarchy of models and modeling approaches such as stochastic models of ion channels and receptors for ligand binding calculations, ordinary differential equation lumped cell models, and partial differential equation continuum models at the tissue and organ levels. It also requires the model parameters at one scale to be linked to detailed models of structure and function at a smaller spatial scale – hence the need for “multi-scale modeling”.

Progress on developing physiome-style models of organ systems, in which physical conservation laws are solved on anatomically-based models, is well underway for the heart, lungs, musculo-skeletal system, digestive system and some sensory organs. In all of these projects the goal is to construct models that incorporate the detailed anatomy and tissue structure of the organs in a way that allows the inclusion of cell-based models and the spatial distribution of protein expression.

The long term challenge for the Physiome Project is to build a modeling framework in which the effect of a gene mutation can be modeled all the way from its effect on protein structure and function to how the altered properties of the protein affect a cellular process such as signal transduction, and how the changed properties of that process alter the function of tissues and organs. There will be many other benefits from this integrative framework. Understanding how model parameters are affected by individual variation, by embryological growth, by ageing and by disease, for example, will bring enormous benefits to the design of medical devices, the diagnosis and treatment of disease and the development of new drugs.

References

  1. Hunter, P.J., Robbins. P. and Noble, D., 2002. The IUPS Human Physiome Project. European Journal of Physiology 445 (1): 1-9.
  2. Hunter, P.J. and Borg, T.K., 2003. Integration from proteins to organs: The Physiome Project. Nature Reviews Molecular and Cell Biology. 4:237-243.
  3. Crampin, E.J., Halstead, M., Hunter, P.J., Nielsen, P.M.F., Noble, D., Smith, N.P.and Tawhai, M., 2004. Computational physiology and the Physiome   Project. Exp. Physiol. 89:1-26.

(I) IUPS is the International Union of Physiological Sciences ( www.iups.org ) and the IUPS Physiome Project is run under the auspices of the IUPS Physiome and Bioengineering Committee, co-chaired by the author and Prof Aleksander Popel.

Links:

back



Pharmacokinetic modeling & Simulation – in drug discovery and early drug development. Where do we stand?

Frank-Peter Theil (Hoffmann-La Roche Ltd., Basel)

A battery of modeling and simulation tools have been developed during the last years, which allow to predict pharmacokinetic key properties solely based upon in silico and in vitro information. On one hand, these tools allow the separate prediction of pharmacokinetic sub-processes such as absorption, distribution and elimination. On the other hand, it is also possible to incorporate that information into a whole body physiologically based pharmacokinetic model. Such models can be used to predict the pharmacokinetics of novel drug candidates in body compartments of interest (plasma, blood, tissues) solely based upon readily available in silico and in vitro information. These novel models are applied for prospective prediction of the human situation with regards to pharmacokinetic/pharmacodynamic behavior of drug candidates as well for mechanistic evaluations in order to understand key properties of clinical drug candidates. The ultimate goal is to trigger a more rationale decision making at early stages of drug discovery and development, which also helps to avoid attrition at later stages of drug development. An update of the current status of the available preclinical modeling and simulation technology will be provided.

[BC]2 presentation

back


 

Bioinformatics applications for food and nutrition: From genomes to biological systems

Frank Desiere (Nestle)

The recent publication of the Human Genome poses the question:  how will genome technologies influence food development? Food products will be very different within the decade with considerable new values added as a result of the biological and chemical data that bioinformatics is rapidly converting to usable knowledge. Bioinformatics will provide details of the molecular basis of human health. The immediate benefits of this information will be to extend our understanding of the role of food in the health and well-being of consumers.

In the future, bioinformatics will impact foods at a more profound level, defining the physical, structural and biological properties of food commodities leading to foods with greater quality in all aspects. Bioinformatics will improve the toxicological assessment of foods making them even safer. Eventually, bioinformatics will extend the already existing trend of personalized choice in the food marketplace to enable consumers to match their food product choices with their own personal health.

To build this new knowledge and to take full advantage of these tools there is a need for a paradigm shift in assessing, collecting and sharing databases, in developing new integrative models of biological structure and function, in standardized experimental methods, in data integration and storage, and in analytical and visualization tools. Examples will be given from genome analysis, transcriptome experiments and proteome identification, finally leading to a systems-biology view of human-nutrition.

[Presentation not available]

back

 

 

 

ADRIS - A New in silico Approach to Collect, Store, Mine, Simulate, and Visualize Pharmaco- and Toxicogenetic Information Related to Severe Adverse Drug Reactions

Joseph Gut (Therastrat)

J. Gut, D. Bagatto, R. Boeuf, R. Dannecker, H. Hug, S. Saxer, R. Schindler, A. Stephan, and A. Wess.
TheraSTrat AG, 4123 Allschwil, Switzerland.

Severe adverse drug reactions (ADRs) occur unpredictably and at low frequency upon exposure of human individuals to drugs posing a major problem to drug development, clinical practice, safety assessment and regulation of drugs. Recent meta-analysis by Lazarou et. al. (JAMA, 279:1200-05) suggests that in the USA alone, ADRs are responsible for about 100’000 fatalities per year, making them the fourth to sixth leading cause of death. ADR-related annual health care costs are estimated at approx. US$200 billion.


Our current understanding of basic pharmaco- and toxicokinetic parameters allows us to associate certain ADRs with well characterized genetic variations, but many ADRs still remain unpredictable. They often show a marked interindividual susceptibility and can be mediated by toxic metabolites and their interactions with cellular targets. Even though we have long surpassed single gene analysis from an experimental point-of-view (e.g. high-throughput SNP analysis), no tools exist to put the acquired data in the context of the structural diversity of chemical compounds that can cause ADRs, as well as in the context of many other critical parameters such as genomic, proteomic, epigenetic and metabolic factors. Data and knowledge on risk factors related to ADRs is dispersed in numerous databases, publications, and other resources in mostly non-compatible formats and a vast amount of new data relevant for the susceptibility of individuals to ADRs is emerging rapidly.


Here, we present a novel approach to conceptionally organize and logically, as well as semantically, relate heterogeneous data. This approach, termed Adverse Drug Reactions Information Scheme (ADRIS), integrates “theragenomic” data such as structures of parent compounds, metabolites, metabolic pathways, covalent adducts, pharmaco-, toxico-kinetic and -dynamic data, protein sequence and structure information, genetic variations, gene expression data, as well as toxicological and clinical outcome data for individual patients. ADRIS constitutes the basis for the in silico knowledge management and discovery system SafeBase™.


SafeBase™ presents the user with a holistic view on ADRs and assists in the identification of risk factors responsible for predispositions of ADRs. It provides a novel approach to collect, store, mine, simulate, and, most importantly, visualize theragenomic data, which is rapidly becoming a critical component in modern drug design and development and in the assessment of drug safety and efficacy, with the goal to develop safer and more individualized medicines. In this presentation, the concept and the implementation of ADRIS in SafeBase™ will be presented and illustrated with a number of examples how new pharmacogenetic knowledge is created through the integration of theragenomic data.

[BC]2 presentation

back
 

 

 

Modelling repolarization and re-entrant arrhythmia.

Denis Noble (University Laboratory of Physiology, Oxford, UK)

The long duration of the cardiac action potential is made energy-efficient by the very low permeability of the i K1 channel at plateau potentials (Noble, 2002a, b) . The price paid is that repolarization is fragile and depends on the highly-reactive channel i Kr – a reactivity that causes many problems for the development of successful new drugs. Modelling the repolarization process and its spatial distribution is now well-advanced and can successfully reproduce the form of the T wave (Antzelevitch et al. , 2001) and so contribute to understanding QT problems. To complete this understanding and to make it successfully predictive we need

• to extend modelling to human cells,   

• to ensure accurate reconstruction of excitability changes during and following repolarization,

• to incorporate these models into tissue and organ models of re-entrant arrhythmias.

I will describe some new work (Ten Tusscher et al. , 2003) that achieves some of these aims and which could form the basis for solving the QT problem.

  1. Antzelevitch, C. , Nesterenko, V. V. , Muzikant, A. L. , Rice, J. J. , Chien, G. & Colatsky, T. (2001). Influence of transmural gradients on the electrophysiology and pharmacology of ventricular myocardium. Cellular basis for the Brugada and long-QT syndromes. Philosophical Transactions of the Royal Society, series A 359, 1201-1216.
  2. Noble, D. (2002a). The heart cell in silico: successes, failures and prospects. In In Silico Simulation of biological processes , vol. 247. ed. Novartis-Foundation , pp. 182-197. Wiley, London.
  3. Noble, D. (2002b). Modelling the heart: insights, failures and progress. BioEssays 24, 1155-1163.
  4. Ten Tusscher, K. H. W. J. , Noble, D. , Noble, P. J. & Panfilov, A. V. (2003). A model of the human ventricular myocyte. American Journal of Physiology , 10.1152/ajpheart.00794.02003.

[BC]2 presentation

back

 

 



Applications of Virtual Screening using High-Throughput Docking

Eric Vangrevelinghe (Novartis Institutes for BioMedical Research )

High-Throughput Docking (HTD) is now routinely used for in-silico screening of compound libraries and for selecting hit candidates which most likely bind to the active site of a target. Although this technique requires knowledge of the 3D structure of the therapeutic target, there are many possible cases in which HTD can contribute to the drug discovery process. A review on several different applications of HTD at Novartis will be given.

[BC]2 presentation

back


 

 

Support Vector Machines – an example of supervised learning in bioinformatics.

Guido Steiner ( F. Hoffmann-La Roche Ltd.)

 

Analyses of highly multivariate data sets, especially in “High Dimension - Low Sample Size” situations (e.g. microarray experiments), play an increasingly important role in bioinformatics and often drive classical statistics to its limits. Support Vector Machines (SVMs) is one of the newer approaches that try to address these problems, with the potential of delivering reliable classification rules even in the case of very high dimensional and noisy data sets where overfitting becomes a significant problem. The method is characterized by a solid theoretical basis and has already performed well in different areas of bioinformatics. As a supervised approach, SVMs detect informative patterns in a given set of training samples with the aim of generalizing on new, previously unseen cases. This talk provides a basic introduction into the ideas of statistical learning and shows some real-life application examples of SVMs in biomedical research.

[BC]2 presentation

back



Inferring Principles of Regulatory Design from high-throughput biological data

Erik van Nimwegen (Biozentrum, University Basel)

In recent years, high throughput technologies have provided a wealth of genome-wide data that allow investigations of the global functional and regulatory organization of cells. In this talk I will discuss recent work in which I used comparative genomic and expression data to infer principles of organismal regulatory and functional design. Topics that will be discussed include: genome-wide discovery of bacterial regulons through comparative genomics; genome-wide discovery of developmental enhancers; functional organization of mRNA decay rates in the cell; and scaling in the functional gene-content of organisms.

[BC]2 presentation

back


 


From Data Mining to Biomolecular Simulations @ IBM Research

Alessandro Curioni (IBM Research Zurich)

An overview will be presented of the activities in Computational Biology at IBM Research.  A particular emphasis will be given to molecular simulations and the way they can complement the huge amount of biological information currently available.

 

Links:

back



Towards Quantitative Biology: The Development of Integrated Computational Systems to Understand Disease Pathways and to Guide the Discovery of Novel Therapeutics

Hans-Peter Fischer (Genedata)

 

Through the availability of the complete genome sequences of human and a range of important eukaryotic model organisms (e.g., worm, fruit fly,   mouse and rat) and the development of large-scale experimental techniques (e.g. transcriptomics, proteomics), we have entered a new era in   biomedical research. For the first time ever, researchers have the technological means to gather comprehensive data on basic biological phenomena and diseases mechanisms, as well as the   possibility to monitor the effect of drug candidates on a molecular   level.

Our ultimate goal is to develop quantitative models that can be used to predict the effects of a therapeutic interference, based on the chemical structure, the patients genetic makeup and the   given environmental conditions. It is expected that such quantitative   models of diseases will revolutionize the drug discovery process, as they will enable a directed approach to the discovery and modification of bioactive molecules as potential therapeutic agents. Potential drug safety issues could be identified at a very early stage by simulating the effect of the drug candidate on the human body. Using   such a virtual research environment scientists can simulate large-scale experiments in silico that could take months or years to do   in the lab or clinic.

The current bottleneck for such an approach lies in the complexity of the human organism's “blueprint” on the cellular, tissue and organ level, as well as the multitude of cross-talk channels between these levels. Here, we present Phylosopher™, an integrated computational platform that combines complete genome sequences of the human genome together with the genomes of relevant model organisms. Phylosopher™ provides the framework for integrating biological data (including   physiologic and environmental) in the context of a disease, with a focus on understanding and determining clinical responses to potential treatment. For this, the system integrates comprehensive data on the encoded genes, transcripts, splicing variants, promoters and SNPs, respectively, along with proteomic data, metabolic and signalling pathways, and protein-protein interaction data.

We will show how the integrated handling of genome, transcriptome, proteome and metabolome data facilitates the construction of quantitative models for the human organism. Additionally, we will present how genetic regulatory networks can be reconstructed, signalling pathways can be elucidated and how this knowledge can be used in pharmaceutical applications. The examples will cover applications in target identification and validation, assay development, and compound mode-of-action evaluation for various indications, including oncology, inflammatory diseases and infectious diseases.

 

back


 

Swiss-Prot: the challenges of extracting and representing protein information

Amos Bairoch (Swiss-Prot group, Swiss Institute of Bioinformatics)


In the context of the UniProt [1] project (see www.uniprot.org ), the Swiss-Prot knowledgebase [2] strives to maintain and expand its role in providing to the user community, a high-quality and up to date resource on proteins. Such a mission requires that we respond both to the evolving needs of our users and to the tremendous increase in sequence data which is the hallmark of modern biomolecular research activities. It is therefore important to increase our productivity in terms of information extraction and to develop new representations of the data tailored to the specific requirements of some user communities.

In our presentation, we will briefly discuss four developments that address some of these concerns:
 

  •  The Anabelle annotation tool: how to extract more efficiently information from protein sequences using state of the art sequence analysis software.
  •  How can we progress from the manual and time-consuming approach of extracting information from publications to a supervised text-mining driven mode;
  •  The representation of human disease mutations and polymorphisms at the level of the protein sequences in Swiss-Prot;
  •  NiceProt: a tool to explore the content of Swiss-Prot entries.
      

[1] Apweiler R., Bairoch A., Wu C.H., Barker W.C., Boeckmann B., Ferro S., Gasteiger E., Huang H., Lopez R., Magrane M., Martin M.J., Natale D.A., O'Donovan C., Redaschi N., Yeh L.S. UniProt: the Universal Protein knowledgebase, Nucleic Acids Res. 32:D115-D119(2004).

[2] Bairoch A., Boeckmann B., Ferro S., Gasteiger E. Swiss-Prot: juggling between evolution and stability, Briefings Bioinform. 5:1-17(2004).

[BC]2 presentation

back



Mechanistic Systems Biology Modeling Applied to the Pre-Clinical Cardiac Safety Assessment of a Pharmaceutical Compound: From Channels to Cells to Tissue

Gabriel Helmlinger (Novartis Pharmaceuticals, CH)

A. Georgieva, G. Helmlinger (Novartis Pharmaceuticals, USA), D. Bottino, S. Lett (The BioAnalytics Group LLC, USA), C. Penland (Predix Pharmaceuticals, USA), A. Stamps (Dept. of Chemical Engineering, University of South Carolina, USA)

This presentation will focus on an integrative, mechanistic systems modeling approach to cardiac electrophysiology. Via an integrated suite of models from channel to cellular and ultimately tissue levels, we used specific compound data from limited pre-clinical cardiac electrophysiological studies, such as channel assay as well as one action potential assay, to:

(i) estimate missing compound data at the channel level through the application of reverse-engineering techniques;

(ii) integrate measured and estimated channel values into action potentials at the cellular level;

(iii) further integrate such cellular responses to predict transmural ventricular responses, which in effect represent ECG-like features at the tissue level, such as QT prolongation, transmural dispersion of repolarization.

The combination of experimental data and computer-based predictions of compound-induced changes at every level of cardiac biology organization (channel, cell, tissue) provides valuable, quantifiable information to aid in the pre-clinical cardiac safety assessment of new compounds.

[BC]2 presentation

back


 

Handling Complexity When Simulating Biological Systems

Jürgen Klenk (Definiens AG)

According to Physics Nobel Laureate Gerd Binnig, creativity and cognition are deeply related processes. In particular, these processes can be considered evolutionary (Binnig calls this “Fractal Darwinism”). In this talk I would like to share lessons learned from designing a “Cognition Network” model and building “Cognition Machines”, which can drive cognitive and thus creative processes, and the application of Cognition Machines to the simulation of complex systems such as Biological Systems. Definiens has built and markets two Cognition Machines, for automated understanding of and knowledge extraction from complex images (Cellenger®), and text documents (Polymind®). These two Cognition Machines cover the collection of the first ingredient to simulation – data. The second ingredient – the dynamic model – could be expressed as a Cognition Network model, into which the collected data could be automatically integrated. The resulting “Simulation Cognition Machine” would then be capable of performing the third step – carrying out the simulation process whilst properly handling the system's complexity due to its fractal structure.

[BC]2 presentation

back

 



Configurable & Integrated Life Science Informatics: Putting Data to Work

Mark Demesmaeker (Spotfire AB)

The constant evolution in life science research technologies dramatically impacts half-life of bio- and cheminformatics applications. Thus, the ability to quickly modify and adapt informatics platforms remains a key factor of success. DecisionSite Guided Analytics makes it possible for informatics specialists to capture the analysis process as it is conducted and rapidly turn around and make that configured process available to other scientists. Processes, methods, and the Guided Analytics that support them may be developed iteratively in a very collaborative environment, which allows life science companies create repeatable, sharable, expert-developed data analysis procedures and routines. Taking advantage of integrating with 3rd party-applications or open-source statistical environments such as R researchers may incorporate any desired function or algorithm.

[BC]2 presentation

back


 

In silico analysis of eukaryotic promoters.

Edward J. Oakeley (Friedrich Miescher Institute)

Edward J. Oakeley, Alessandro Di Cara, Karsten Schmidt

We have developed an automated analysis tool for the visualisation and analysis of eukaryotic promoters in a high throughput manner. The user starts with a list of genes obtained from expression analysis using Affymetrix microarrays, he/she then runs the Affymetrix IDs through a promoter extraction tool which uses a database of experimentally verified and predicted starts of transcription as a source (Genomatix GmbH, Germany). Promoters that have not been mapped by Genomatix are guessed by comparison with mapped orthologues in other species using the EBI DBA Block aligner programme. Promoters are automatically scanned for potential transcription factor binding sites using the Transfac 7.4 matrices (Transfac, Germany) and similarity patterns between the promoters are identified and displayed in a web page output in SVG format.

[BC]2 presentation

back


Simebac, a bacterial metabolism simulation

Eric Fourmentin-Guilbert (Fondation Fourmentin-Guilbert)

With the progress of information technology, it is conceivable that, within the next 15 years, we will be able to integrate detailed biochemical information into a single data model representing all molecules inside a cell. Today, we can already start by building elementary building blocks of simple biological functions.

The SIMEBAC project aims at developing a 3D molecular model of a bacteria (E. coli). The first objective is to model a simple function such as the DNA transcription. The role of the Foundation is to foster the start-up of the project through different actions like the development of a network of excellence, workshops, and prizes.

More detailed information is available at www.fourmentinguilbert.org.

[BC]2 presentation

back


 

Information extraction in Molecular Biology and Biomedicine.

Alfonso Valencia ( National Center for Biotechnology, Madrid)

Integrating vast amounts of heterogeneous data, generated by Genomics and Proteomics approaches, poses serious challenges in Bioinformatics and Computational Biology. In particular there is an emerging need of combining new experimental information with the one accumulated in scientific publications and other text repository (see Valencia, EMBO Reports, 2002).

The main challenges for the development of Information Extraction and Text Mining in Molecular Biology and Biomedicine can be summarized as: i) access and organization of the textual information (document retrieval); ii) development of comprehensive repositories of annotated text in the various knowledge domains; iii) identification of entities in text, particularly protein and gene names, but also diseases, drugs, species, tissues and others; iv) accurate description of the relations between entities at the level of pair wise relations (e.g. protein interactions and gene control relations), relations between entities (e.g. genes associated to a given disease), and at the level global relations (e.g. function common to a set of genes?), and v) representation of the extracted knowledge, including technical issues (e.g. graphical formats, database querying capabilities) and scoring and summarizing the information extracted from text. For a review see, Blaschke et al., Brief. in Bioinfor. 2002.

Even if these problems are common to other areas of Information Extraction, the highly specialized nature of this area of research makes necessary the development of specific tools and applications. One clear example is the specific need of building explicit connections between the facts stored in domain databases, e.g. annotations in the Swissprot database, to the underlying knowledge archived in scientific publications.

In response to these challenges it is very important to compare and assess competing approaches with common standards and evaluation criteria (see the description of the BioCreative competition).

In this presentation I'll review the situation of the field, the possibilities offered by current systems to end-users, and my view on the impact in these developments in development of Biotechnology and Biomedicine.

Links:

[BC]2 presentation

back

 


Elaborate pores and complex machines: nature's nanotechnology benchmarks

Helmut Grubmüller (MPI for biophysical Chemistry,Göttingen)

Selectivity and Efficiency of Water Channels

Bert L. de Groot and Helmut Grubmüller

Theoretical and Computational Biophysics Department, MPI for biophysical Chemistry,Göttingen

Aquaporins are highly selective water channels. Molecular dynamics simulations of multiple water permeation events correctly predict the measured rate and explain why these membrane channels are so efficient, while blocking other small molecules, ions, and, in particular, protons.

[BC]2 presentation

back


Biozentrum, University of Basel