High Performance Computing

The success of the drug discovery process is now directly related to a company's computation capabilities. Results and research projects that were unimaginable a few years ago are accessible with today's supercomputers. Industry sources expect the IT life-science market to explode to more than $9 billion by 2003. Major IT companies have made significant investments and formed partnerships, with IBM and Compaq each committing $100 million. Data overload from automation and robust database technology means that companies have an immense amount of data on both drug targets and lead compounds. The downside is that these companies are not equipped in infrastructure or organization to take full advantage efficiently of this gold mine. There are also questions of staff: Do we have the right people? How do we use our existing in-house bioinformatics team? Should we outsource or build our own? What's a fair contract for services, profit sharing, and leasing? How many teraflops are enough? This meeting will seek to cover these issues for biotech, pharma, software, and hardware engineers, for application specialists, and for anyone who envisions the marriage of supercomputers and the life sciences.

Advisors and Chairs
Dr. Philip E. Bourne, University of California, San Diego and Keck Graduate Institute and The Burnham Institute
Mr. Peter Ungaro, IBM Corporation
Dr. Fred Wright, The Ohio State University
Dr. Siamak Zadeh, Sun Microsystems, Inc.

Additional Speakers
Mr. Lars Bauerle, Spotfire Inc.
Dr. Denise Beusen, Tripos, Inc.
Dr. William J. Camp, Sandia National Laboratories
Dr. George Davidson, Sandia National Laboratories
Mr. Martin Gollery, TimeLogic Corporation
Mr. Andrew Grimshaw, Avaki Corporation
Dr. Ralf Gruber, Lausanne (EPFL), Switzerland
Mr. Joshua J. Harr, Linux NetworX, Inc.
Dr. Sorin Istrail, Celera Genomics
Dr. Tod M. Klingler, Structural GenomiX, Inc.
Mr. Avery Moon, Infotone Communications Corporation
Dr. Arthur J. Olson, The Scripps Research Institute
Dr. Peter Rieger, Kelman GmbH
Mr. Yury Rozenman, Platform Computing Corporation
Dr. Mark Swindells, Inpharmatica
Dr. Margaret Werner-Washburne, University of New Mexico
Dr. Lisa Yan, Accelrys
Dr. Bo Yuan, The Ohio State University

Systems
High-Performance Computing: Providing the Foundation to Meet the Data and Compute Demands of R&D
Grids: Harnessing Data and Computer Resources in a Multisite, Multiorganizational Context
Effective Cluster Management: Making the Most of Your High-Performance Computing Investment
Unlimited Power: The Benefits of Acceleration in Compute Farms
Distributed Processing in the Life Sciences

Architecture and Software
Data Management Challenges in the Post-genomics Era
Terascale Computing and the New Biology: Embarrassingly Parallel Informatics and Massively Parallel Simulation
Information Interaction for eAnalytic Applications: Delivering Information to End-Users in a Distributed Information Architecture
Assembly and Annotation of the Entire Human Genome Using Large Linux Clusters
Customizing the Design and Configuration of Network and Hardware Components to Support Specific Applications
Pharmaceutical Informatics: Pharmacophore Identification Using Massive Distributed Computing

Applications: Data Mining
From First Assembly towards a New Cyberpharmaceutical Computing Paradigm
Specialized Integration Hardware Facilitating High-Performance Automated Systems Integration and Simultaneous Analysis of Heterogeneous Bioinformatic Data
High-Performance Functional Annotation of Proteomic Sequences Using a Distributed Computing System
High-Performance Computing: Parsing the Human Genome
Visual Comparison of Multiple, Genome-Scale Data Sets in Yeast: Insights into Cellular Processes
Computational Challenges of High-Throughput Technologies

Applications: Modeling and Simulation
Parallel Pipelines from Genes to Putative Drug Targets
Structure-Based Drug Design on a Genomic Scale
Biopendium™: How We Use 1100 Commodity Processors to Precalculate Relationships between a Million Protein Sequences and Do It Again and Again
Internet Distributed Docking: FightAIDS@home
Protein Pathway Profiling

7:00am Registration, Poster and Exhibit Viewing, and Light Continental Breakfast

8:00 Chairperson's Opening Comments
Mr. Peter Ungaro, Vice President, High Performance Computing, Life Sciences, IBM Server Group, IBM Corporation

8:10 High-Performance Computing: Providing the Foundation to Meet the Data and Compute Demands of R&D
Mr. Peter Ungaro
Information technology has become an integral part of many biotech efforts today. IBM is working with many companies on building an IT framework for life sciences including areas such as high-performance computing and data integration. This presentation will center around IBM's experiences in this area and show some examples of how technology is being used in leading-edge life-science companies today.

8:40 Grids: Harnessing Data and Computer Resources in a Multisite, Multiorganizational Context
Mr. Andrew Grimshaw, Chief Technical Officer and Founder, Avaki Corporation
This presentation will begin with a brief introduction to grid computing-an exciting outgrowth of the national supercomputing centers. Next the conversation will focus on the use of data grids to provide secure access to remote data at both collaborator sites and remote sites within the same organization. The access methodologies and the security environment will both be presented in detail. The use and benefits of computational grids will then be presented, along with specific examples from the life-science arena. The presentation will conclude with discussion of the future of grids in life sciences.

9:10 Effective Cluster Management: Making the Most of Your High-Performance Computing Investment
Mr. Joshua J. Harr, Linux NetworX, Inc.
A large cluster is a serious investment, yet far too often cluster purchases are made without serious thought as to how to keep the cluster usable and available to the users. This talk will lay the groundwork for effective cluster management. I will discuss the common challenges of managing a large cluster such as how to maintain software consistency and how to handle upgrades, how to identify and manage hardware problems, and how to effectively monitor and control activity in the cluster. These issues and others will be approached from a best-practices approach with the intent that the attendee will leave with a better understanding of how to deal with the major issues in cluster management.

10:30 Unlimited Power: The Benefits of Acceleration in Compute Farms
Mr. Martin Gollery, Director of Bioinformatics Research, TimeLogic Corporation
Server farms provide a great advantage in price performance over traditional supercomputers. For the types of truly enormous analyses that are necessary today, however, server farms have to grow to enormous size. Accelerated server farms are hundreds to thousands of times faster than CPU-only farms, reducing manpower and electrical and floor-space requirements and enabling entirely new types of analyses. A comparative analysis will be provided using several popular bioinformatics algorithms on a number of genomes.

11:00 Distributed Processing in the Life Sciences
Mr. Yury Rozenman, Business Development Manager, Life Sciences, Platform Computing Corporation
Distributed computing played a significant role in accelerating the sequencing and assembly of the human genome-ahead of the original schedule. Homologies between sequences, as well as matches to Markov models and results of several applications, are all computed, worldwide, by automatic workflow systems. Gene expression, proteomics, and large-scale genotyping are deployed and their results deposited in standardized repositories. With system and life-science software vendors, Platform Computing has acknowledged this requirement and is addressing the identified challenges through key integrations that incorporate distributed computing technologies seamlessly and deeply into the discovery process. Using technology from Platform Computing, scientists can easily exploit diverse data sources and multiple analysis tools without worrying about such details as program or data location, data interoperability, and communication. Through illustrative examples based on real-world experiences involving platform solutions, the demand for distributed computing can be effectively addressed.

1:45 Chairperson's Comments
Dr. Siamak Zadeh, Group Manager, Life Sciences, Sun Microsystems, Inc.

1:50 Data Management Challenges in the Post-genomics Era
Dr. Siamak Zadeh
Following the mapping of the human genome, the life-science community is faced with a new challenge; having access to gene sequences is merely the first step toward using the vast volume of genomic data to advance biological research and drug discovery. In the post-genomics era, researchers will need to access, analyze, and mine vast volumes and multiple types of data on a grand scale, with scientists around the globe using information from myriad sources, including microarray expression analysis, SNP analysis, and other tools for locating genetic variation, transgenic comparisons, and volumes of associated clinical and biological data. In this presentation, some of the underlying challenges and efforts to address these will be discussed.

2:20 Terascale Computing and the New Biology: Embarrassingly Parallel Informatics and Massively Parallel Simulation
Dr. William J. Camp, Director, Computation, Computers, and Mathematics, Sandia National Laboratories
Terascale computing is a foundational technology of genomic biology. Proteomics appears to require 10 to 100 times as much computing power, and optimal target identification together with design of intervention may require petascale computing. Assembling the genome involves only large-scale informatics-essentially a communication-free embarrassingly parallel process. The classification of protein expression in signaling and metabolic pathways of diseased versus healthy tissues will also be an informatics problem, albeit one of hugely greater complexity. When we turn to understanding the roles of various proteins in those pathways, we shall almost certainly need computationally intense simulation. This shift in emphasis only increases when we deal with targets, small molecule intervention strategies, and toxicity mitigation. Thus, informatics and simulation pose quite different architectural requirements on computing; informatics involves almost no floating point arithmetic, needs fairly minimal communications capabilities, and tends to be strongly input-output bound. By contrast, simulation and optimization are highly dependent on floating point operations, usually involve less input-output, and require an effective communications fabric among processors. With this in mind, I will discuss the likely roles of compute farms, Beowulf clusters, scalable clusters, and massively parallel supercomputers in the new biology.

2:50 Information Interaction for eAnalytic Applications: Delivering Information to End-Users in a Distributed Information Architecture
Mr. Lars Bauerle, Product Manager, Spotfire Inc.
Advances in information technology and computing have allowed companies to collect more data from their research, product development, manufacturing, and distribution activities than ever before. Increasing amounts of data, with decision making being pushed down to people throughout the value chain, have made for extremely complex information architectures at many organizations. Analysis of data and sharing the results have to be performed faster and with more accuracy to support critical decisions in all facets of a competitive business. The ability to deliver information to analysis applications from disparate data sources seamlessly and simultaneously is a basic requirement for a productive analysis environment that enables users to make informed decisions. This presentation will cover Spotfire's approach to providing end-users with information through a combination of configurable and role-based information interaction applications that allow a diverse community to access, transform, and interact with data across multiple domains and disciplines. Examples of how major pharmaceutical and biotechnology companies have already successfully integrated the Spotfire approach in their genomics, screening, and other data environments will also be discussed.

4:00 Assembly and Annotation of the Entire Human Genome Using Large Linux Clusters
Dr. Bo Yuan, Head, Bioinformatics Group, Human Cancer Genetics Program, The Ohio State University
The recent draft assembly of the human genome provides a unified basis for describing genomic structure and function. The draft is sufficiently accurate to provide useful annotation, enabling direct observations of previously inferred biological phenomena. We report here a functionally annotated human gene index placed directly on the genome. The index is based on the integration of public transcript, protein, and mapping information, supplemented with computational prediction. We describe numerous global features of the genome and examine the relationship of various genetic maps with the assembly. In addition, initial sequence analysis reveals highly ordered chromosomal landscapes associated with paralogous gene clusters and distinct functional compartments. Finally, these annotation data were combined to produce observations of gene density and number that accord well with historical estimates. We estimate that the genome contains 65,000 to 75,000 transcriptional units, with exon sequences comprising 4%. The creation of a comprehensive gene index requires the synthesis of all available computational and experimental evidence.

4:30 Customizing the Design and Configuration of Network and Hardware Components to Support Specific Applications
Dr. Ralf Gruber, Director CAPA, Swiss Federal Institute of Technology, Lausanne (EPFL), Switzerland
Up to now, application software engineers had to adapt their programs to the architecture of the parallel machine that computer vendors designed. With the arrival of computer vendor independent cluster computers, the network and computing architecture can now be tailored to the specific needs of the applications. Such a cost-effective parallel machine is well balanced in terms of processor performance, communication network speed, and, last but not least, storage and archiving requirements. To guarantee a most efficient job scheduling with no data flow bottlenecks, the customized IT installation is subdivided in a front-end and a compute farm. The front-end takes care of all the data-handling tasks such as data and job preparation, data storage and archiving, database access, and job scheduling. The farm is a cluster in which the different compute nodes are interconnected by a network that can consist of high-speed crossbar components such as Myrinet or Quadrics or just of a low-cost and low-speed Fast Ethernet. This application-tailored concept has successfully been applied to a proteomic factory including a farm of over 1,400 Alpha processors.

5:30 Reception and Tour of The San Diego Supercomputer Center (sponsored by Cambridge Healthtech Institute)

Join Us for an Opening Night Reception and
Tour of the San Diego Supercomputer Center!

SDSC - a unit of UC San Diego
San Diego Supercomputer Center

We have arranged with the San Diego Supercomputer Center (SDSC) to offer the attendees of CHI's High-Performance Computing meeting the opportunity to view this state-of-the-art center. SDSC's 250 staff members have expertise in computational science research and the development and integration of high-performance computing technologies. The scientific staff specializes in biology, biomedicine, bioinformatics, chemistry, and environmental sciences. Technology researchers focus on scalable parallel and distributed computing, data-intensive computing, applied network research, scientific visualization, and security. Support staff develop and maintain systems and application software, provide researcher and operations support, conduct training, develop user documentation, and publish information about current projects. SDSC supports three supercomputers for the national research community: IBM RS/6000 SP and CRAY T3E parallel supercomputers and a CRAY T90 vector supercomputer. These systems are available through a peer-review system to academic researchers and students in the United States.
The evening will commence at 6:00 p.m., Monday, January 14, with an opening night reception at the Center, followed by an introductory talk by Dr. Philip Bourne, after which the 45-minute tour will begin. The SDSC is a short distance from the hotel and shuttles will be provided. Please indicate on the registration form if you are interested in attending. For questions regarding the tour, please contact Elizabeth Cutler by e-mail at ecutler@healthtech.com.

8:30 Chairperson's Comments
Dr. Fred Wright, Assistant Professor, Human Cancer Genetics, The Ohio State University

8:35 From First Assembly towards a New Cyberpharmaceutical Computing Paradigm
Dr. Sorin Istrail, Senior Director, Informatics Research, Celera Genomics
The new science of whole genomics will fundamentally change how pharmaceutical companies pursue the vital challenge of new and better drugs. Target discovery, lead compound identification, pharmacology, toxicology, and clinical trials are likely to merge with the science of bioinformatics into a powerful system for developing new pharmaceutical agents. It will be possible to simulate the action of new molecules or therapeutic programs against diverse metabolic pathways prior to preclinical testing. Thus, a paradigm of cyberpharmaceutical testing will be available to the industry, speeding the selection of promising new agents, eliminating products that are likely to exhibit toxicity, and reducing the formidable costs and risks associated with the current paradigm of drug development. We will report on Celera's design of a whole genome shotgun assembler and its application to the sequencing of the Drosophila and human genomes. We will also present some of the major emerging computational challenges of the above paradigm in the exciting new areas of proteomics, structural genomics, expression profiling, SNPs, and pharmacogenomics.

9:05 Specialized Integration Hardware Facilitating High-Performance Automated Systems Integration and Simultaneous Analysis of Heterogeneous Bioinformatic Data
Mr. Avery Moon, Chief Executive Officer, Infotone Communications Corporation
Integrating heterogeneous, discrete databases and making them collectively available to analytic software are challenges constraining the effective aggregation, interpretation, and analysis of bioinformatic data. Traditional software integration techniques are proving ill suited for solving this problem as they require extensive custom programming, suffer performance issues, and require highly specialized training. This presentation will cover emerging hardware-based integration techniques that eliminate such challenges by using high-performance cluster-based data integration algorithms, dedicated-purpose logical interface cards, and flexible integration expansion modules to provide highly scalable multicomputing of databases and computational algorithms on a single platform.

9:35 High-Performance Functional Annotation of Proteomic Sequences Using a Distributed Computing System
Dr. Lisa Yan, Senior Manager, Protein Engineering, Accelrys
Genome sequencing projects have produced a vast amount of data in recent years. High-performance computing systems have become an essential part of biocomputing as efficient tools for mining these data. We introduced GeneAtlas™, a software pipeline for high-throughput functional annotation of protein sequences. A typical GeneAtlas run consists of a large set of independent tasks and achieves its top performance on both distributed computing systems and multiprocessor supercomputers. We will discuss the configuration of an in-house Beowulf cluster and the GeneAtlas performance based on the case study on available human proteomic sequences.

10:05 High-Performance Computing: Parsing the Human Genome
Dr. Fred Wright
The availability of the human genome draft sequence is enabling the exploration of a host of important biological phenomena, including the distribution of genes, the nature of sequence identity, and other aspects of chromosomal structure. Many of these analyses are conceptually simple but require substantial computation. We discuss efforts to address these problems using high-performance computing clusters.

11:15 Visual Comparison of Multiple, Genome-Scale Data Sets in Yeast: Insights into Cellular Processes
Dr. Margaret Werner-Washburne, University of New Mexico, and Dr. George Davidson, Sandia National Laboratories
Genomics and the ability to measure the "state" of a cell or tissue through expression profiling have laid the groundwork for a revolution in understanding cellular function. As more and different types of data become available, it is increasingly clear that new tools are needed not only to handle the data and to extract information but also to allow scientists from disparate areas to analyze biological systems in light of this information. The sheer numbers of datapoints in genome-scale data make analysis of single data sets a daunting task. We are applying an ordination and visualization program developed at Sandia National Laboratories, VxInsight, to allow concurrent visualization and analysis of different gene-expression data sets as well as different types of genome-scale data. Through this analysis, we have been able to provide novel support for identification of protein function, differentiation of cellular function, and identification of potentially novel drug targets.

11:45 Computational Challenges of High-Throughput Technologies
Dr. Denise Beusen, Discovery Software Product Manager, Tripos, Inc.
In the last decade, pharmaceutical companies under pressure to sustain their profitability have embraced high-throughput technologies to enhance research productivity. There is now a realization that the resulting volumes of data create challenges that can offset the anticipated rewards, with no net gain in shortening the pathway to a new drug. These challenges include ways to store and search chemical structures within relational databases, compact representations for storing millions of chemical structures, viewing and analyzing high-dimensional data, and integrating data for delivery to disparate elements of an organization. This presentation will focus on computational tools and strategies that address these challenges and enable knowledge-driven research decisions.

1:45 Chairperson's Comments
Dr. Philip E. Bourne, Professor, Department of Pharmacology, and Senior Principal Scientist, San Diego Supercomputer Center, University of California, San Diego; PDB Project Team Leader and Adjunct Professor, Keck Graduate Institute and The Burnham Institute

1:50 Parallel Pipelines from Genes to Putative Drug Targets
Dr. Philip E. Bourne
Data-driven biology is a prime candidate for high-performance computing. I will describe our efforts to establish a processing pipeline on three systems: the IBM Blue Horizon 1500-node RS/6000 SP machine, a Compaq cluster of 96 nodes running Linux, and the machines composing the national grid. Our starting point is a set of 20,000 identified genes for which we wish to define putative protein 3-D models. I will take a pragmatic approach considering the successes and failures as we try and get real work done on a large scale.

2:20 Structure-Based Drug Design on a Genomic Scale
Dr. Tod M. Klingler, Vice President, Structural GenomiX, Inc.
One of the most promising strategies for using genomic information to improve drug development is through a combination of structural genomics and structure-based drug design. Recent advances in computational modeling techniques and the industrialization of experimental structure determination now allow the production of structural information for almost half of all known proteins. At Structural GenomiX we are integrating experimental and computational approaches to produce the most comprehensive and accurate view of protein structure space and building methods for using this view for directing the drug development process. In this talk I will describe this integration of technologies and its application in drug design.

2:50 Biopendium™: How We Use 1100 Commodity Processors to Precalculate Relationships between a Million Protein Sequences and Do It Again and Again
Dr. Mark Swindells, Chief Scientific Officer, Inpharmatica
Inpharmatica's Biopendium™ product is the world's leading database of precalculated protein structure predictions. We will discuss our experience of developing and managing an 1100 processor compute farm to produce Biopendium™ and what we see as the issues for commercial throughput computing in bioinformatics. One key aspect often overlooked is the management skills needed for the transition from prototype to production, and we will talk about some of the challenges our organization has successfully overcome.

3:50 Internet Distributed Docking: FightAIDS@home
Dr. Arthur J. Olson, Professor, Department of Molecular Biology, The Scripps Research Institute
We have developed a new computational approach to simulating the evolution of drug resistance and screening drug candidates in silico for robustness in the face of resistance mutations. The approach is based upon the use of co-evolutionary algorithms to test and rank competing populations of trial drugs and mutant drug targets. Our latest effort involves extending these approaches with a fully detailed atomic model by using our AutoDock flexible drug docking in a massively parallel computational environment. In collaboration with Entropia, we have established an Internet-based community (currently over 25,000 individuals) that has donated its computers to run our docking computations on HIV-protease drug screening (FightAids@Home). With the orders of magnitude more computing power, we anticipate the development of extended novel methodologies, as well as greatly improved modeling results. This talk will focus on our recent experience and results with an Internet-based computing platform.

4:20 Protein Pathway Profiling
Dr. Peter Rieger, Head, Methods Engineering, Kelman GmbH
Investigation of single components is no longer sufficient in genome and proteome research; today's challenge is to make gene functionality transparent from complex gene-to-gene interrelationships by combining methodically independent approaches. Here we demonstrate a systematic IT-based approach for in silico protein-protein interaction mapping, applicable to complete genomes. Functional gene networks produced by SPRAB™ (Selective Protein Recognition And Binding) technology reveal the complete set of putative interactions between all proteome constituents known. This study focuses on the yeast genome as a model to estimate the computational costs the entire human genome would require, if one considers all molecular versions gene products exist in. We will show how the computational approach copes with the problem of data complexity and volume during preparation, calculation, and analysis. In parallel, IT solutions for visualization of and navigation in the whole complex space of gene-to-gene relationships are presented that enable the user to include information gained otherwise, such as experimental data of interactions, expression data, etc. Thus, a new dimension of uncovering and profiling protein pathways can be presented. The approach of considering and combining different aspects of gene interactions allows a raster search for relevant pathways, providing an in-depth understanding of gene function.

Corporate Sponsor Biography

Avaki Corporation is a provider of software infrastructure allowing applications to run securely across multiple enterprise domains, geographically diverse sites, and distributed networks. AVAKI is providing solutions that form the basis for web services to a wide range of industries, including Life Sciences, Financial Services, and Engineering Intensive Manufacturing. For more information, please visit www.avaki.com.

IBM is the world's largest information technology company, with 80 years of leadership in helping businesses innovate. IBM Life Sciences brings together IBM resources, from research and e-business expertise to data and storage management and high-performance computing, to offer new solutions for the life sciences market, including biotechnology, genomic, e-health, pharmaceutical, and agri-science industries. The fastest way to get more information about IBM is through the IBM home page at http://www.ibm.com.

Platform Computing is the leading software provider for distributed computing infrastructure, providing technology that enables enterprises to optimize existing computing resources. Founded in 1992, Platform has over 1,000 customers worldwide and works closely with major system vendors such as Compaq, HP, IBM, SGI and Sun, that bundle, distribute and co-market Platform solutions to their customers. The company is headquartered in Toronto, Ontario, with 400 employees at 13 offices around the world. For more information visit www.platform.com

Hotel Information
January 14-15, 2002
The Marriott La Jolla
4240 La Jolla Village Drive
La Jolla, CA 92037
Tel: 858-587-1414
Fax: 858-546-8518
Cut-off Date: December 21, 2001
Room Rate: $169 S/D

Please call the hotel directly to make your room reservation. Identify yourself as a Cambridge Healthtech Institute conference attendee to receive the reduced room rate. Reservations made after the cut-off date or after the group room block has been filled (whichever comes first) will be accepted on a space-and-rate-availability basis. Rooms are limited, so please book early.

Travel Information
Special Airline Discounts Available
Special zone and discount fares have been established on United Airlines for this conference. Please call United Airlines Meeting Reservations Center directly at 800-521-4041. You must reference ID #579YS.

Call for Posters
Cambridge Healthtech Institute encourages attendees to gain further exposure by presenting their work in the poster sessions. Please fill out the registration form, with the poster title and primary author. To ensure inclusion in the conference binder, a one-page summary must be submitted and registration must be paid in full by December 14, 2001. POSTER INSTRUCTIONS

Call for Exhibitors
This conference will attract scientific researchers, executives, managers and lab directors from pharmaceutical and biotechnology industries, as well as academic and research institutes. Companies with services or products related to, high performance computing, distributed computing, database integration, bioinformatics, protein folding, drug screening analysis, and other technologies and services related to computing will greatly benefit by exhibiting. Since the exhibition space at this event is extremely limited, booths are only available to sponsors of this event. For more information on sponsorship, please contact Jim MacNeil at 781-972-5441 or jmacneil@healthtech.com

Corporate Sponsor

Associate Sponsor

Sponsoring publications Bioinform Drug Discovery and Development Genome Technology Genomics and Proteomics Scientific Computing and Instrumentation Technology Review