spacer
EBIInformation

Therapeutic Applications of Computational Biology - Posters

4th - 6th September, 2005

The following posters will be presented at the conference:
  • Poster 1, EBIMed: Text Mining services at the EBI
  • Poster 2, Development of a safety intelligence network
  • Poster 3, Reactome: a Database of Biological Pathways
  • Poster 4, Integr8: an integrated bioinformatics portal for complete genomes and proteomes
  • Poster 5, Bioinformatics tools for annotation of protein sequences in structure-based drug discovery
  • Poster 6, The Gene Ontology Annotation (GOA) Database: sharing biology with GO as an aid to biomedical research and pharmaceutical discovery
  • Poster 7, A computational platform to determine kinase inhibitor mechanism of action
  • Poster 8, The PRIDE database: providing proteomics data to inform the medicine of tomorrow
  • Poster 9, Chemical Entities of Biological Interest
  • Poster 10, Connection In silico between Medicinal Plants and Animal Venoms
  • Poster 11, Prioritizing genomic drug targets in pathogens
  • Poster 12, Optimizing library design for resequencing by hybridization
  • Poster 13, Networking for the transcriptional regulation of diabetes using literature mining techniques
  • Poster 14, Bioisosterism and molecular recognition in the biological space. A case study
  • Poster 15, Interaction of WR99210 with dihydrofolate reductase thymidylate synthase (DHFR-TS) From Plasmodium vivax
  • Poster 16, Environment-specific amino acid substitutions in membrane proteins: towards more effective homology recognition and comparative modelling for key drug targets
  • Poster 17, Comparative genome analysis applied to drug development
  • Poster 18, BioSapiens: a European Network for integrated genome annotation
  • Poster 19, Bioinformatics identification of tolerance markers in organ transplantation
  • Poster 20, TRANSFOG: translational and functional onco-genomics
  • Poster 21, How to make analyses of high-throughput experiments reproducible and extensible?
  • Poster 22, Toxicogenomics resources standards, ArrayExpress infrastructure and datasets
  • Poster 23, Modelling human signaling pathways for complex diseases
  • Poster 24, Not so many drug targets after all? Reviewing the evidence for a low human protein-coding gene number
  • Poster 25, ChemaPhore: tool for pragmatic computer-aided structure-based drug design
  • Poster 26, The IntAct Database
  • Poster 27, The important role of calcium in regulation of adhesion disassembly and cell migration: mathematical modelling
  • Poster 28, Target drugability assessment
  • Poster 29, Fragment-based screening
  • Poster 30, Pathways to Discovery: Leveraging pathways to gain novel insights into biology for drug discovery, development and clinical studies

Presenting authors are highlighted in red bold.


Poster 1
EBIMed: Text Mining services at the EBI

Miguel Arregui, Harald Kirsch, Sylvain Gaudan and Dietrich Rebholz-Schuhmann
EMBL–European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge CB2 6SA, UK
EBIMed is the EBI’s new web service that combines document retrieval and information extraction from Medline. It performs keyword searches to retrieve Medline abstracts, which then are analyzed for associations between biomedical terms, in particular UniProt protein and gene names, GO annotations, drug names and species. The identified concepts are displayed in a table and offer access to key sentences and the original abstracts. All discovered concepts are linked to the EBI’s and NCBI’s biomedical databases for further inspection. EBIMed’s text analysis profits from a set of state-of-the-art text-mining modules, specially optimized for high-throughput text analysis. EBIMed runs on a Linux cluster that is part of EBI’s computational services. EBIMed is designed to support molecular biologists who intend to overview Medline abstracts instead of reading them all one by one. EBIMed is available to all users world-wide via any Web browser free of charge. http://www.ebi.ac.uk/Rebholz-srv/ebimed.
Top
Poster 2
Development of a safety intelligence network

Julie C. Barnes
BioWisdom, Harston Mill, Harston, Cambridge, UK
Intelligence networks are large-scale representations of knowledge, derived from a diverse array of sources and ‘codified’ to be semantically consistent and highly navigable. The network format provides huge gains in information transparency and can accommodate the inevitable changes in context and perspective that occur across the drug development process. This poster will describe the development of a safety intelligence network that is being applied to address business issues relating to drug-induced toxicity. As an example, we will describe how the safety intelligence network has been applied, using only publicly available information, to uncover aspects of the aetiology, pathology and molecular/cellular mechanisms involved in drug-induced muscle toxicity. An example of how scientifically logical workflows can call on the ‘codified’ knowledge to identify biomarker candidates for drug safety will also be presented.
Top
Poster 3
Reactome: a Database of Biological Pathways

Ewan Birney1, Bernard de Bono1, David Croft1, Peter DEustachio2, Marc Gillespie2, Gopal Gopinathrao2, Bijay Jassal1, Geeta Joshi-Tope2, Suzanna Lewis2, Lisa Matthews2, Esther Schmidt1, Lincoln D. Stein2 , Imre Vastrik1 and Guanming Wu2 1EMBL–European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge CB2 6SA, UK
2Cold Spring Harbor Laboratory, 1 Bungtown Road, Cold Spring Harbor, New York 11724, USA
Biological processes are complex, self-regulated, feedback controlled networks. Reactome describes life processes in a top-down fashion that is useful for didactic and data retrieval purposes. The primary concepts in Reactome are events and entities, and the data is represented using six main types: reaction, pathway, macromolecule, small molecule, complex, and catalyst activity. Data is cross-referenced to many existing web-based informatics resources. The Reactome web interface gives a textbook-like view of cellular processes for biologically oriented users, while a series of query tools allow bioinformaticists to search the database and make discoveries. The data content of Reactome as well as its software and data model are freely reusable under an open source license. Reactome is available at http://www.reactome.org.
Top
Poster 4
Integr8: an integrated bioinformatics portal for complete genomes and proteomes

Lawrence Bower, Alan D. Horne, Paul J. Kersey and Rolf Apweiler
EMBL–European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge CB2 6SA, UK
Integr8 is a new web portal for exploring the biology of organisms with completely deciphered genomes. For over 250 species, Integr8 provides access to general information, recent publications, and a detailed statistical overview of the genome and proteome of the organism constructed using data resources such as the UniProt Knowledgebase, Genome Reviews, IPI, InterPro, CluSTr and GOA. Users can customise their own interactive analysis through the BioMart application, and can download both customised and pre-prepared datasets for their own use. Integr8 can also be used to identify putative orthologues and paralogues and to explore potential regions of synteny between genomes. Integr8 is available at http://www.ebi.ac.uk/integr8.
Top
Poster 5
Bioinformatics tools for annotation of protein sequences in structure-based drug discovery

Suzanne C Brewerton
Astex Therapeutics, 436 Cambridge Science Park, Milton Road Cambridge CB4 0QA, UK
Structural biology techniques such as X-ray crystallography and NMR spectroscopy are essential parts of the Astex Therapeutics integrated discovery engine, Pyramid™. Structural methods allow identification and characterization of drug fragments bound to target proteins such that they can be rapidly transformed, using efficient medicinal chemistry and structure-based design approaches, into potent, selective lead compounds. One of the great challenges of structure-based drug discovery is that most regulatory proteins in man, the obvious targets for new drugs, are complex multi-domain proteins. Computational techniques can provide tools for the prediction of structural features in proteins, such as domains, a priori and by detection of similarity to known structural motifs. This enables the laboratory researcher to make decisions before carrying out a time-consuming series of experiments to which the protein is not amenable. Several tools have been designed for use in construct design for protein expression, purification and crystallization.
Top
Poster 6
The Gene Ontology Annotation (GOA) Database: sharing biology with GO as an aid to biomedical research and pharmaceutical discovery

Evelyn Camon, Daniel Barrell EBI, Emily Dimmer EBI, Nicky Mulder EBI and Rolf Apweiler EBI
EMBL–European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge CB2 6SA, UK
The number of large-scale experimental datasets generated from high-throughput technologies has grown rapidly. Finding disease-related genes requires laborious examination of hundreds of possible genes many lacking in functional characterization. Now, more than ever the organizing and sharing of ‘known’ biological knowledge across organisms and databases is vital. Biological knowledge resources such as the Gene Ontology Annotation (GOA) database (http://www.ebi.ac.uk/GOA), which provides high-quality functional annotation to proteins within the UniProt Knowledgebase and plays an important role in database interoperability and knowledge integration. The integration of GOA with analytical tools has aided the clustering, annotation and biological interpretation of such large datasets which can help in the identification of candidate ‘disease’ genes. Furthermore GOA is also useful in the development and validation of automated annotation tools, in particular text-mining systems. The increasing interest in GOA highlights the great potential of this freely available resource to assist both the biomedical research and bioinformatics communities.
Top
Poster 7
A computational platform to determine kinase inhibitor mechanism of action

Bruce Church, Andrew Stubbs, Sergej V. Aksenov, Yi Zhou, Anjali Dihman, Jorge Vialard, Tim Perera, and Iya Khalil.
Johnson & Johnson, PRDBE, Beers, Belgium
A major challenge in drug development is discovering the molecular mechanism of action (MOA), and the efficacy and safety for new compounds. A common approach towards gaining biological insights into drug MOA action is to determine statistically significant changes in molecular expression and to overlay these changes onto known pathways. However most profiling techniques identify co-regulating gene transcripts and do not reveal causal relationships, and known pathways may not be relevant in determining the compound’s effects on disease progression. A more powerful approach is to discover the causal regulatory pathways that the drug affects (both known and unknown) directly from the ‘omic’ response data. Here we present a reverse engineering method that can determine the regulatory pathways affected by specific perturbations within the context of the cell or organism from ‘omic’ profiling data. We applied our approach to determine the regulatory pathways affected by known kinase inhibitors. Our approach led to the identification of gene regulatory pathways that differentiated between kinase inhibitors in cell lines. This can also be extended to analyzing ‘omic’ data from clinical samples providing a framework for preclinical to clinical translation of mechanism of compound action.
Top
Poster 8
The PRIDE database: providing proteomics data to inform the medicine of tomorrow

Richard Cote1, Lennart Martens2, Phil Jones1, Chris Taylor1, Henning Hermjakob1 and Rolf Apweiler1
1EMBL–European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge CB2 6SA, UK
2Dept Medical Protein Research, Flanders Interuniversity Institute for Biotechnology, Faculty of Medicine and Health Sciences, Ghent University, Rommelaere Institute - Building D, A. Baertsoenkaai 3, B-9000 Ghent, Belgium
The volume of mass-spectrometry derived protein identifications is increasing exponentially. Unfortunately, access to this potential goldmine of information is often limited to published digested data. The raw data, if available, is usually spread across small project-centric databases in incompatible formats. To alleviate these issues, we created the Proteomics Identifications Database (PRIDE): a centralized, standards compliant, public data repository. PRIDE implements data schemas and controlled vocabularies brought forward by the Proteomics Standards Initiative (PSI), such as mzData for peak lists and mzIdent for search engine results. With the wealth of information in PRIDE, data mining approaches could extract potential new drug targets and make predictions on isolating proteins or protein families for use as potential biomarkers. The Human Plasma Proteome Project (HPPP) has already deposited its findings in PRIDE, with the Human Brain Proteome Project scheduled to submit follow later this year. It is our conviction that PRIDE can become an invaluable tool in biomedical research, as well as for the proteomics community at large.
Top
Poster 9
Chemical Entities of Biological Interest

Kirill Degtyarenko, Paula de Matos, Marcus Ennis and Martin Zbinden
EMBL–European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge CB2 6SA, UK
ChEBI is a freely available dictionary of chemical compounds, with IUPAC and NC-IUBMB-endorsed terminology. ChEBI’s focus has been on nomenclature but it also incorporates a chemical ontology and cross-references to UniProt. Compounds contained in ChEBI are either the products of nature or are synthetic products used to intervene in the processes of living organisms. Intervention may be purposeful or accidental, e.g. chemicals in the environment. Molecules not directly encoded by the genome are excluded (nucleic acids, proteins and peptides derived from proteins by cleavage). ChEBI is a curated database. Every compound has been assigned an official ChEBI name. An IUPAC semi-systematic name has also been added where available. Relevant synonyms, CAS numbers and formula are also included. Where these have been taken from external sources the cross references to these data sources are displayed. In future releases, ChEBI hopes to provide structural information. ChEBI is available at http://www.ebi.ac.uk/chebi/.
Top
Poster 10
Connection In silico between Medicinal Plants and Animal Venoms

Silvana Giuliatti1, Andreimar M. Soares2 Thais J. Pereira3, Camila Prates3, Renato Puga3, Marcela C. de Santo3 and Milton Faria Jr3
1FMRP - University of São Saulo Amuí
2FFCLRP - University of São Paulo
3UNAERP - University of Ribeirão Preto
Introduction: Among the great variety of medicinal plants there is a specific kind, the antipoisonous medicinal plants. Populations of diverse countries live with the danger of poisonous animals. Bioinformatics systems help the storage and analysis of large quantities of data. The aim of this project is the development of a system and a database for the integration of antipoisonous medicinal plants and animal venoms. Methodology: We used the LINUX-RedHat operating system, a MySQL database and Perl programming language. Results: Protein sequences of animal poisons had been obtained through available databases, such as the NCBI. Information on antipoisonous medicinal plants and their properties have also been stored in a database. The interactions between animal toxins and plant inhibitors have been analyzed. The analysis results are available to the academic public. Conclusion: The present project has shown to be a potential tool for analysis and development of new drugs.
Top
Poster 11
Prioritizing genomic drug targets in pathogens

Samiul Hasan and Mark J. Schreiber
The Novartis Institute for Tropical Diseases, 10 Biopolis Road, #05-01 Chromos, Singapore 138670
The target identification stage represents the first step in the drug discovery process. A suitable drug target must satisfy a number of factors to permit progression to the next stage. We have developed software for weighting and integrating desired properties on every gene in a pathogen to enable ranking them as drug targets. The application is written generically to allow users to define new drug target-ranking features and adjust the confidence placed in a specific feature. We have tested this approach on the genome of Mycobacterium tuberculosis.
Top
Poster 12
Optimizing library design for resequencing by hybridization

Niall J. Haslam1, Nava E. Whiteford1, Gerald Weber1, Jonathan Essex1, Cameron Neylon1 and Adam Prügel-Bennett2
1School of Chemistry, University of Southampton
2School of Electronics and Computer Science, University of Southampton
High-throughput sequence analysis methods offer the promise of viral resequencing as a clinical diagnostics and pharmacogenomics tool in which the viral sequence will guide the therapeutic strategy. Current sequencing technology is limited in terms of quantity of information and cost. Many groups are working on high-throughput sequence analysis mostly focused on shorter read lengths. We present a method of library design for the resequencing of small viruses and other small regions of genomic information that is optimized for short read lengths. This type of library requires a different design approach to gene expression arrays. We will highlight some of the issues by selecting a number of examples of current targets in pharmacogenomics. These examples will demonstrate the advantages and shortcomings of our system. Although this work is in progress, our preliminary results indicate that it should be possible to use smaller numbers of shorter probes in libraries for resequencing.
Top
Poster 13
Networking for the transcriptional regulation of diabetes using literature mining techniques

Sigeo Ihara, Naoko S. Nishikawa, Jun’ichi Kitakami, Patrick C. Reid, Singo Tsuji, Shogo Yamamoto and Hiroyuki Aburatani
Laboratory for Systems Biology and Medicine, Research Center for Advanced Science and Technology, #70, The University of Tokyo, 4-6-1 Komaba, Meguro, Tokyo 153-8904 Japan
To identify target genes from gene expression results such as microarray analysis, the efficient utilization of the knowledge of protein–protein, as well as protein–DNA interactions stored in MEDLINE is useful. We have developed a system called APISST (Advanced Protein Interaction Search SysTem) by combining a literature mining method using natural language processing with graph theoretic analysis. Previously we have applied our method to the expression data sets of microarray analysis for cancer and cardiovascular disease. Here we applied our method to the microarray data for diabetes reported by H. Thomas et al. By calculating the number of interaction edges for each protein node within the interaction networks obtained by APISST, the importance of genes of interest in pathways is determined. The common pathways and the specific pathways related to different patterns of target genes in INS-1 cells are obtained. Our results suggest that our method is useful in analyzing diabetes.
Top
Poster 14
Bioisosterism and molecular recognition in the biological space. A case study

Antonio Macchiarulo and Roberto Pellicciari
Dip. Chimica e Tecnologia del Farmaco, via del liceo 1, Perugia 06100, Italy
There is growing awareness that the translation of the ever-increasing number of lead compounds to clinical candidates is still a very slow and often inefficient process [C.Booth et al. Nat.Drug.Discov. 2004]. To facilitate the lead optimization procedure, due consideration must be given to the use of right bioisosteric replacements. In this work, the concept of bioisosterism is revisited using a holistic approach. As a case study, the bioisosteric relationship between three different acidic groups, carboxylate, sulfonate, and phosphonate, is investigated by looking at the differences in the properties of their binding sites. These properties were used to define a biological space composed of constellations of binding sites. The differences in the occupation of diversity space by binding sites of acidic groups provides a better understanding of the concept of bioisosterism and its biological relativeness. The biological space, defined in this work, complements the previously definitions of chemical space.
Top
Poster 15
Interaction of WR99210 with dihydrofolate reductase thymidylate synthase (DHFR-TS) From Plasmodium vivax

Ankit K. Misra, Prashanth K. Aitha, Meenakshi Malhotra, Vidya Kothekar and Tannistha Nandi
Department of Biotechnology, Jaypee Institute Of Information Technology, Noida, (U.P) 201307 India
Dihydrofolate reductase thymidylate synthase (DHFR-TS) has been studied as a target molecule to combat P. falciparum malaria. Development of drug resistance due to fast mutations is reducing the efficiency of antifolates as antimalarials. This phenomenon has been linked to the occurrence of mutations in the parasite's DHFR-TS. Mutations at residues 51, 59, 108 and 164 have shown to be linked with resistance of P. falciparum to antifolate antimalarials. N51 and I164 are conserved in both P. falciparum and P. vivax. However, C59 and S108 are substituted by S58 and T118. By studying the known interaction of WR99210 with P. falciparum we have identified the conserved contact residues and repeated the study for PvDHFR-TS with the same ligand. WR99210 was found to interact with conserved residues of PvDHFR-TS and thus in this study we tried to design a ligand, such that resistance is not developed. Ternary complexes of various inhibitors (modified ligand) and the PvDHFR-TS have been studied to identify the contact regions and best-optimized structure complex using Hyperchem 7.5. Mutant inhibitors were screened using distance matrix algorithm and the most suitable ligand conformation is proposed. By lead optimization, new antimalarial drugs can be found that do not develop resistance.
Top
Poster 16
Environment-specific amino acid substitutions in membrane proteins: towards more effective homology recognition and comparative modelling for key drug targets

Younes Mokrab and Kenji Mizuguchi
Department of Biochemistry, University of Cambridge, Cambridge, UK
Comparative/homology modelling for protein structure prediction can help overcome the significant lack of structural information for transmembrane (TM) proteins. Our work aims at developing a sequence–structure homology recognition method that uses environment-specific substitution tables (ESSTs) and structure-dependent gap penalties to (1) increase the accuracy of alignments involving TM protein sequences and structures and (2) improve the specificity and sensitivity of homology searches for TM proteins. A comprehensive set of homologous families of highly curated alignments comprising high-resolution structures and close sequence homologues was constructed, containing 945 helical TM protein domains. The derived ESSTs were used to obtain new rules about patterns of amino acid substitutions in the various lipid bilayer compartments. A series of ESSTs are optimised to improve alignments and sequence-structure homology recognition involving TM proteins.
Top
Poster 17
Comparative genome analysis applied to drug development

Jean-Marc Neefs, Peter Verhasselt, Peter Leemans and Bruno Neys Johnson&Johnson Pharmaceutical Research&Development
R207910 is a diarylquinoline drug active against Mycobacterium tuberculosis. To identify drug resistance mutations and possibly understand the drug mechanism of action, complete genome sequencing of five different drug-sensitive and drug-resistant strains of M. tuberculosis and M. smegmatis was performed using different technologies. We used BLAST to map raw sequences to genomes of each species, and dedicated software to record differences between sensitive and resistant strains. This approach rapidly led to identification of about 50 putative mutations. Potential mutated genes were identified by adapting several standard gene prediction methods. Experimental validation reduced the original number to less than 10 mutations. ATPE, the gene coding for ATP synthase F0 subunit, was mutated in all resistant strains. Transfection of a drug-sensitive M. smegmatis strain with the mutated F0 operon caused drug resistance at levels similar to mutant strains. This further led to the hypothesis for the mode of action of R207910.
Top
Poster 18
BioSapiens: a European Network for integrated genome annotation

The BioSapiens Consortium
www.biosapiens.info
BioSapiens is a European network for integrated genome annotation, funded by the EU’s 6th Framework Programme, and made up of bioinformatics researchers from 25 institutions in 14 countries. The objective is to provide a large-scale, concerted effort in genome annotation, using both informatics tools and input from experimentalists.
Top
Poster 19
Bioinformatics identification of tolerance markers in organ transplantation

Paul Perco1, Peter Blaha2, Alexander Kainz1, Bernd Mayer3, Peter Hauser1, Thomas Wekerle2 and Rainer Oberbauer1
1Department of Nephrology, Medical University of Vienna, Austria
2Division of Transplantation, Department of Surgery, Medical University of
Vienna, Austria
3emergentec biodevelopment, Vienna, Austria
Tolerance induction by mixed chimerism and costimulation blockade is a promising approach along organ transplantation. Such procedures might reduce the risk for organ rejection also under reduced or omitted immunosuppressive medication. We present a bioinformatics procedure comparing a group of bone marrow transplanted mice treated with an anti-CD40L antibody and a group of bone marrow transplanted mice without the co-stimulation blockade. The bioinformatics workflow applied included statistical and explorative methods extracting a core differential gene expression profile. Next this profile was expanded using co-regulation analysis via identification of enriched promoter motifs, as well as by extended protein network analysis. This procedure allowed, next to a purely statistical analysis, also a biological interpretation of differential gene expression results. The tight integration of gene expression experiments and bioinformatics resulted in a list of about 100 genes presumably associated with tolerance. These results will be subsequently tested in validation experiments.
Top
Poster 20
TRANSFOG: translational and functional onco-genomics

Antony F. Quinn
EMBL–European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge CB2 6SA, UK
The TRANSFOG project aims at the systematic identification and functional characterization of novel cancer genes with high potential diagnostic and therapeutic value in breast, colon and lung cancer. An important aspect of the project is the dynamic integration of a variety of experimental and clinical data from diverse sources, including transcriptomics, epigenomics, proteomics and functional assays. To this end, the Distributed Annotation System protocol (R. Dowell et al., BMC Bioinformatics 2001, 2, 7) and Dazzle server framework (http://www.biojava.org/dazzle/) have been extended to provide ontology-based gene and protein annotation suitable for comprehensive analysis by both humans and machines.
Top
Poster 21
How to make analyses of high-throughput experiments reproducible and extensible?

Markus Ruschhaupt1, Wolfgang Huber2, Annemarie Poustka1, Ulrich Mansmann3
1Deutsches Krebsforschungszentrum, Im Neuenheimer Feld 580, 69120 Heidelberg
Germany
2EMBL–European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge CB2 6SA, UK
3Institut für medizinische Informationsverarbeitung, Biometrie und Epidemiologie
Marchioninistr. 15, 81377 München, Germany
The analysis of data from high-throughput technologies involves a combination of different processing steps, each of which may consist of a complex bioinformatic or statistical algorithm. Reproducibility and extensibility of results is a crucial, but difficult task – even if the raw data has been made public. We use the tool of a compendium to achieve reproducible calculations and to offer an extensible computational framework. A compendium is an interactive document that bundles primary data, processing methods (computational code), derived data, and statistical output with the textual documentation and conclusions. Our compendia are based on the vignette and packaging technology available from the R and Bioconductor projects. An example can be downloaded from http://www.bepress.com/sagmb/vol3/iss1/art37.
Top
Poster 22
Toxicogenomics resources standards, ArrayExpress infrastructure and datasets

Susanna A. Sansone, Philippe Rocca-Serra, Sergio Contrino, Mohamad Shojatalab, Ugis Sarkans, Niran Abeygunawardena, Alvis Brazma and the Microarray Informatics Team
EMBL–European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge CB2 6SA, UK
Toxicogenomics adds microarray technology to the field of toxicology, as a new tool to predict and elucidate mechanisms of toxicity. However, technology acceptance by regulatory bodies requires thorough validation via phenotypic anchoring. Creating tools for data submission and review is a therefore a critical step. We report here the extension of ArrayExpress’s infrastructure to support toxicogenomics data and metadata. The successful upload of the complete ILSI-HESI dataset via Tox-Miamexpress, a ‘Tox’ extension of the ArrayExpress web-based deposition tool endorsing MIAME/Tox specifications, and additional datasets using dedicated spreadsheets have demonstrated that MAGE can be harnessed to encapsulate both transcriptomic and classical toxicology (pathology, clinical chemistry) data. Access wise, the ArrayExpress infrastructure now boasts a sophisticated data warehouse, enabling combined microarray/tox endpoints queries to be performed. Finally, tight collaboration with US counterparts (CEBS at NIEHS and NCTR-FDA) is ongoing in an effort to establish the basis of an international infrastructure for Toxicogenomics data.
Link: http://www.ebi.ac.uk/microarray/Projects/tox-nutri/index.html.
Top
Poster 23
Modelling human signaling pathways for complex diseases

B. Sonneborn1, A. Schüler1, S. Perrey1, H. Brinck1 and T. Wiebringhaus1,2
1Bioinformatics and Mathematics, University of Applied Science of Gelsenkirchen, Recklinghausen
2Biopinio, Recklinghausen
Signal transduction is an important biological process by which cells transduce an external signal into a cellular response. To model signalling events for target identification, we have compiled a manually curated protein–protein interaction (PPI) network from ~15,000 human PPIs. We have calculated network parameters, e.g. diameter, the number and size of hubs and the scale-free behaviour in dependence of the network size. We propose a signalling pathway as a series of PPIs between a plasma membrane receptor and a transcription factor and restricted the path to a sequence of three subcellular compartments required for the transduction of the signal. We computed paths for different path lengths and calculated the number of transcription factors that are reachable per receptor in dependence of the path length and present the relevance of hubs in this context. To evaluate our approach we mapped differentially expressed genes from complex diseases onto this pathways and show first results.
Top
Poster 24
Not so many drug targets after all? Reviewing the evidence for a low human protein-coding gene number

Christopher Southan
Molecular Pharmacolgy, AstraZeneca, Mölndal, Sweden and University of Nottingham
Estimates of human protein-coding genes, an important parameter for assessing potential drug target numbers, have oscillated wildly. This review assesses data sources pertinent to the basal (unspliced) count. Despite increased transcript coverage, a finished assembly and refinements in automated annotation, the Ensembl count of 22218 is 1828 fewer genes than in 2001. UniGene and UniProt suggest redundancy-reduced human mRNA and protein collections are flattening out at ~ 28000 but high-throughput submissions include artefactual ORFs. Database surveys suggest genuine novel protein discovery has slowed to a trickle and do not support the hypothesis that substantial numbers of short proteins remain undetected in mammals. Completed chromosome annotations suggest that many transcripts may not be protein-coding, and MS-based proteomics has not verified any unpredicted proteins. The pharmaceutical industry is now faced with the paradox of finite target numbers but increasing protein diversity from SNPs and splicing.
Top
Poster 25
ChemaPhore: tool for pragmatic computer-aided structure-based drug design

Michelle L. Styles, Jun Zeng and Herbert R. Treutlein, Computational Chemistry & Biology Group, Cytopia Research Pty Ltd, Melbourne, Australia
Computer-aided structure-based drug design (SBDD) is now routinely used in the biotechnological and pharmaceutical industries. However, available software packages for SBDD generally don't focus enough on the practical application of these tools in these industries. Many solutions disregard, for example, conformational flexibility of a protein target and available experimental data. We have developed a set of software tools, ChemaPhore, to aid the design of potent and selective inhibitors based on virtual screening of 3D protein structures; binding mode analysis; medicinal chemistry and wet screening. Chemaphore adapts to specific targets, achieving higher productivity than standardized tools. ChemaPhore has been used successfully at Cytopia to design several highly potent and selective tyrosine kinase inhibitors that have entered clinical testing. Benchmarks based on publicly available data have confirmed the high standard of ChemaPhore. An outline of ChemaPhore, including examples from current public benchmark tests and in-house applications will be presented.
Top
Poster 26
The IntAct Database

D. Thorneycroft, K. Robbe, J. Khadake, S. Orchard, S. Mudali, S.Kerrien, C. Leroy, H. Hermjakob and R. Apweiler
EMBL–European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge CB2 6SA, UK
It is becoming clear that most proteins in a given cell are connected through extensive dynamic networks. Improved understanding of these networks can provide information to identify network components as useful drug targets or diagnostic markers. IntAct provides a freely available, open source database system and analysis tools for protein interaction data. IntAct contains data on more than 50,000 protein–protein interactions from a wide range of organisms. IntAct currently hosts numerous interactions covering cancers, mental disorders and host–pathogen interactions involving bacterial and viral infections. Some examples of areas covered are breast cancer with interactions involving the BRCA2 protein, the fmr protein in Fragile-X syndrome, protein networks in Alzheimer’s and Huntington’s diseases and host–viral protein interactions involved in hepatitis. IntAct also contains data from many large-scale studies carried out in model organisms of orthologues to human proteins. All IntAct software, data and controlled vocabularies are freely available at http://www.ebi.ac.uk/intact.
Top
Poster 27
The important role of calcium in regulation of adhesion disassembly and cell migration: mathematical modelling

Najl Valeyev1, Andrei Skorinkin2, Kristy Downing1, Iain Campbell1 and Nicolai Kotov2
1Department of Biochemistry, University of Oxford, Oxford, UK
2Biophysics & Bionics Lab, Department of Physics, Kazan State University, Russia
Cell migration is crucial for many functions in metozoan organisms such as embryonic development, wound repair, cancer invasion and immunity. Migration involves a variety of stages, the most important ones include actin-driven membrane protrusion, new focal adhesion formation and focal adhesion disassembly. Ca2+ dynamics appears to be a crucial factor that regulates the rate of migration. The Ca2+-dependent enzyme, calpain, localizes to and disassembles focal adhesions. Here we develop a model for Ca2+ dynamics and Ca2+-dependent mu-calpain focal adhesion disassembly. We propose that the diffusion-limited Ca2+ entry into filopodia might be a mechanism for cell polarization. The model of the Ca2+ system allows the study of what effects inhibition or activation of various targets on cell migration.
Top
Poster 28
Target drugability assessment

Ursula Egner, Roman Hillig and Martina Schaefer
Schering AG, Research Center Europe, Muellerstr. 178, 13342 Berlin, Germany
In the pharmaceutical industry, the identification and validation of new drug targets has changed fundamentally during recent years. Today, the target discovery process is a multi-step effort with contributions from various areas including bioinformatics, structural biology, cell biology, and functional in vivo studies. In this process, target drugability assessment has become an important decision point before the time-consuming and cost-intensive target validation. In order to classify a potential target candidate as a drugable target, we examine each target protein for its potential to selectively bind a low-molecular-weight compound. A pre-requisite for such analysis is the availability of 3D-structural information of the target. In order to evaluate the potential to achieve selectivity, residues contributing to a binding niche are examined for interactions with a ligand and for the sequence conservation of these residues in the protein family analysed. Target assessment in this sense links information from sequence and structure space.
Top
Poster 29
Fragment-based screening

John Barker1, Oliver Barker1, Michael Charlton1, Thomas Hesterkamp2, Osamu Ichihara1, Pierre Ilouga2, Steffen Köhler2, Joachim Kraemer2, Ingo Krause2, Owen Mather1, Sabine Schaerlt2, Wayne Thomas1, Dirk Ullmann2 and Mark Whittaker1
1Evotec, 151 Milton Park, Abingdon, Oxfordshire, OX14 4SD, UK
2Evotec AG, Schnackenburgallee 114, D-22525 Hamburg, Germany
A key advantage of fragment-based screening is the rapid optimization of ligand–target interactions based on simple chemistry templates and structural hypotheses. Using this approach, chemical novelty can be built in from an early stage in a drug discovery programme. A limitation with the method is the identification of weakly active fragment-molecule hits in the first place. This is typically achieved by NMR studies or high- throughput crystallography and usually up to a thousand fragment molecules are assessed. We have recently demonstrated that weakly active fragments (IC50 values 0.5–2 millimolar) can be reliably identified using fluorescence-based biochemical assays performed on the EVOscreen™ platform, incorporating single molecule detection and 1 microlitre/well screening at up to 100,000 datapoints/day. Initial results of fragment based- screening for two targets, PTP1B and Hsp90, will be presented together with structure determinations by X-ray crystallography of protein-fragment complexes.
Top
Poster 30
Pathways to Discovery: Leveraging pathways to gain novel insights into biology for drug discovery, development and clinical studies

Dana A Abramovitz1, Adam S Corner2
1 Ingenuity Systems, 1565 Charleston Road, Mountain View, California
2 Ingenuity Systems, PO Box 776 Hampton Hargate, Peterborough PE7 8BB
As more information is revealed through large-scale “omics” techniques, it is becoming increasingly apparent that genes do not function alone but through complex biological pathways. Unraveling these intricate pathways is essential to understanding biological mechanisms, disease states, and the function of drugs that transform them.

Ingenuity Pathways Analysis 3.0 is one of the most powerful solutions available today to meet this challenge.

We have evaluated a previously published gene expression dataset that profiles human genes in acute lymphoblastic leukemia in response to single and combination drug treatments. Using relationships from the Ingenuity Pathways Knowledge Base, we built custom pathways to answer specific hypotheses and decipher the complex biological mechanisms underlying patient response to the most effective drug treatment, revealing a novel mechanism of action. By elucidating the complex biological pathways affected by this drug treatment, we can apply this information to the discovery and development of new drugs, to patient stratification studies of new and existing drug treatments, and to understanding the underlying biology and the off-target effects.
Top


Contact

Contact Information

If you have any enquiries or problems regarding the symposium please send an e-mail to tacb@ebi.ac.uk.


spacer
spacer