NIGMS "Visions of the Future" Meeting

Location:
Bethesda Holiday Inn Select

Start Date: 9/23/2002 8:00 AM

End Date: 9/24/2002 4:00 PM

Background

In 1998, NIGMS convened a sampling of scientists representing the broad research community relevant to the Institute's mission. NIGMS recruited the advice and opinions of the scientific community to help assign priorities and set research agendas in anticipation of the planned doubling of the NIH budget. The discussions generated many ideas that blossomed into very successful scientific programs. Among these are the "glue grant" program, the complex systems initiatives, the Protein Structure Initiative, and the NIH Pharmacogenetics Research Network. As the 5-year NIH budget-doubling period draws to a close, the Institute thought it would be wise to take stock of progress, resources, and opportunities, and to again convene a meeting of representatives from the scientific community to identify the most important and emerging areas of biomedical research.

Opening the meeting, acting director Dr. Judith H. Greenberg described how rapidly science has moved in the past 5 years, and how remarkable the changes have been in how scientific research is conducted. Dr. Greenberg also referred to NIH director Dr. Elias Zerhouni's recent "road map" meetings assembled to help chart the course for future NIH-sponsored research, and said that NIGMS hopes its plans will mesh well with the bigger mission of NIH. Dr. Greenberg urged the group to think big, and to identify the most exciting questions likely to face the biomedical research enterprise in 2010, 8 years from now and an amount of time purposefully selected to exceed the lifetime of an individual research grant.

Meeting Summary and Major Themes

A recent convergence of events in biomedical research presents the scientific community with new ways to consider the future of biology. These ways may not have been imaginable just a few years ago. Fundamental mechanisms that underlie many cellular processes have been uncovered, and the sequences of more than 100 organisms have been decoded. This progress provides researchers with an opportunity to explore and understand molecular biology at a revolutionary rate and at a much deeper level. In visiting the many areas of current activity and future need in biomedical research, the meeting discussion revolved around several common areas. The major themes may be categorized as follows, with some overlapping content: modeling/complexity, "mathematization" of biology, tool development/access, comparative genomics, and interdisciplinary science. Another topic that pervaded much of the discussion was the need to find ways to encourage more risk-taking and to fund good ideas rather than "safe" ideas.

Modeling/Complexity

Creating predictive models of cells, tissues, and whole organisms--"virtual entities"--is a worthwhile, but futuristic goal. Many unanswered questions about intracellular and intercellular molecular composition and complexity must be solved before investigators can attempt to model cell behavior. Currently, many basic problems in biology cannot be modeled. Research efforts that may advance progress toward creating predictive models include: amassing comprehensive "parts lists" for various experimental systems, understanding rules of biological design, quantitating cell constituents and reaction rates, and training/maintaining a scientific work force competent in computational biology and physics. A stronger emphasis should be placed on systems biology approaches, especially on analyses of how systems fail. Generating "molecular documentaries" of fundamental molecular processes may be a realistic intermediate goal to modeling larger, singular entities.

Biocomplexity should be a major focus of future biomedical research efforts. Statistical mining of large data sets can yield biologically meaningful patterns that may reveal gaps in knowledge and help to formulate new research questions. The development of algorithms to "reason" with messy biological data may enhance progress in understanding biocomplexity. Advances in approaches to measure molecular and system dynamics will be vital to cracking the complexity of living systems. Clarifying the biological unit of study (studying single cells vs. an averaged, heterogeneous population) will be a critical component to the production of reliable data sets.

"Mathematization" of Biology

A fusion of biology with the physical and engineering sciences is needed. Developing mechanisms that encourage and reward mutual, cooperative interactions among mathematicians, physicists, computer scientists, engineers, and biologists will be important for achieving highly significant future advances. Limiting success in this area is that biologists have not effectively communicated the most important biological problems to quantitative scientists. A further hindrance to progress in the observation that the quantitative components of undergraduate and graduate training in biology are currently inadequate.

Tool Development/Access

Technology development is a key ingredient for advancing biomedical research, including pre-symptomatic and molecular diagnosis of disease. Biosensors and physics-based tools are needed to quantitate levels of biomolecules and flux in individual cells and in samples containing mixtures of biological constituents. Engineering and materials science approaches may provide useful methods to build interfaces with biological systems. Modernized imaging resources are needed to enhance detection of both single molecules and large molecular assemblies. Integrated public databases and accessible computational packages should be developed, although there is also a need for manipulatable software that can collect and code data unique to a given experiment. New technology will likely be very expensive, and technology development grants may afford investigators the means to pursue the development of tools to address their research questions. Miniaturized equipment may help to ameliorate high technology costs, as could national or regional development/distribution centers.

Comparative Genomics

Addressing scientific questions across species, exploiting several model and non-model organisms, can be a valuable tool for uncovering biological design rules, relating genotype to phenotype relationships, and understanding the emergence of novel attributes over evolutionary time. Advances in high-throughput methods for large-scale genome analysis are necessary to facilitate comparative genomics research endeavors. Bioinformatics approaches to extract meaning from genome-wide data sets will also be an important component for progress in this area. Biomedical researchers should gain a better understanding of genomic "regulatory codes" (promoters, introns, and other non-coding regions) and epigenetic mechanisms. Sufficient expertise in the field of microbial physiology is severely lacking, despite a prediction that this area is likely to be a critical component of the future of genomics research. Building bridges with clinical investigators may enhance the flow of genomic knowledge and advance understanding of the mechanistic underpinnings of health and disease.

Interdisciplinary Science

Ongoing, daily collaboration between scientists from different disciplines is vital to the success of future biomedical research. Assembling "critical masses" of personnel representing sufficient breadth and depth of varied scientific expertise will be an important solution to tackling complicated problems in biology and facilitating the fusion of information from disparate sources. Collaboration should be encouraged, not discouraged, for investigators at all stages of their careers. While investigators should be encouraged to work together, they should concentrate on retaining their specialist foci. Establishing "corporate," goal-oriented environments may speed progress toward well-defined areas of interest and scientific need.

Introductions and Preliminary Remarks

Dr. Richard A. Young of the Whitehead Institute for Biomedical Research, and meeting chair, began the discussion by charging the group to seize on the opportunity to provide input to NIGMS in pursuing new scientific directions. Young asked each invited participant to describe his or her research focus and scientific visions for 2010.

Dr. George M. Whitesides of Harvard University described his expertise and interests as straddling four areas: chemistry, biology, materials science, and physics. He identified complexity as a critical topic for discussion. Biophysicist Dr. Lukas Tamm of the University of Virginia recognized the importance of understanding proteins in environments with a focus on gaining knowledge about membranes, including the activities and interactions of membrane proteins, processes such as fusion and fission, and lipid regulation. He spoke about the need for advances in NMR, new engineering approaches to reproduce biological substrates, and nanosensors to detect membrane activity, as means to probe membrane structure and function. Dr. Stephen G. Sligar of the University of Illinois described himself as a physical biochemist and called for a union between the study of physics, chemistry, and cellular communication, all of which have dynamics as a common thread. He noted that reductionism can still be useful, if married to systems approaches to analyze molecular assemblies and the dynamics of cellular processes. Molecular virologist Dr. Edward Scolnick , president of Merck Research Laboratories, said that biomedical research is ready for systems biology approaches, especially as applied to the central nervous system. He cited as rate-limiting the extremely high costs of current and emerging technologies, urging NIGMS and NIH to invest in academic infrastructure to permit widespread acquisition of new technologies, including computation.

Dr. Jean E. Schwarzbauer , a cell biologist at Princeton University, described the extracellular matrix (ECM) as her area of research interest. She recognized the need to fully understand the signaling, assembly, and dynamic function of the ECM by identifying all its individual components and through creating testable, synthetic ECMs. Dr. Alejandro Sánchez Alvarado , a developmental biologist at the University of Utah who studies regeneration in planarians, identified the need to determine the process by which novel attributes arise during evolution. He also pointed to the value of combining cell biological, mathematical, and physical approaches to develop algorithms for various biological processes, such as wound healing. He predicted that deviations from these equations will identify gaps in current knowledge. Evolutionary biologist Dr. Margaret Riley of Yale University expressed interest in understanding how microorganisms interact and defend themselves with biological weaponry, as well as how they develop community responses, such as forming biofilms. She looks to comparative genomics, molecular evolution, and evolutionary theory to guide the development of an evolutionary biological perspective for every antibiotic drug. Dr. Riley also encouraged the development and widespread usage of integrated, universal databases, envisioning that future advances will require teams of scientists trained in different disciplines. Cell biologist Dr. Richard I. Morimoto of Northwestern University urged that science should move "back to the organism," to integrate molecular and cellular information into function. He suggested developing approaches to confront complexity by inviting input from a broad range of specialists, including those from engineering, political science, and management science. Dr. Morimoto also highlighted the importance of NIGMS maintaining strong and diverse training programs.

Dr. Mary Lidstrom , a chemical engineer and bacteriologist at the University of Washington, stressed the importance of microbial physiology. Stating that one of the biggest challenges facing biomedical research is understanding variation in biological processes, she expressed the view that experimental population heterogeneity (deriving averaged data from a mixed population of biological units) can be an obstacle to drawing meaningful conclusions about biology in general. She also noted the importance of new methods for detecting single molecules. Dr. Richard A. Lerner , a chemist and president of The Scripps Research Institute, identified the "critical mass problem" as a deterrent to progress. He suggested that solving complicated biological problems requires the efforts of a large, interdisciplinary group consisting of more than just one or two researchers representing each specialty, and added that substantial research infrastructure must be available to such groups. Dr. Lerner also echoed Dr. Scolnick's comments about expensive technology being rate-limiting to scientific progress. Microbiologist Dr. Sydney Kustu of the University of California, Berkeley stated an importance for determining structures of catalytic intermediates and for developing assays to measure the activity of gas channels. She also recognized the utility of measuring metabolite pools and metabolite flux in growing cells, using sensors to detect nutrients and small molecules and to monitor and regulate allosteric changes during various cell conditions. Dr. Judith P. Klinman , a chemist at the University of California, Berkeley, urged considering not only complexity, but also simplicity. She suggested that many efforts to model natural processes are far from becoming a reality. Dr. Klinman advised that structural studies of proteins should focus not only on known, active sites, but also on other areas of proteins, and that dynamics should be a central focus. She thought that both protein evolution and the multi-functionality of proteins were areas ripe for further research.

Neurobiologist Dr. Erich Jarvis of Duke University Medical Center stated that he aims to understand how the brain generates behavior, using vocal learning in birds as a proxy for human language. He noted that new technologies are very expensive, and he criticized current funding mechanisms to individual investigators that do not encourage risk or collaboration. Structural biologist Dr. Stephen C. Harrison of Harvard University presented his vision as being able to create molecular "movies," or "documentaries," using structural biological techniques and single-molecule imaging to document cellular events occurring during a given process, such as how a virus enters a cell. Dr. Mark Gerstein , a bioinformaticist at Yale University, spoke about the value of being able to view and analyze protein-protein interactions on a large scale, then relate the interactions to function. He also suggested that the structure of scientific information ( e.g. , databases) and scientific publication ( e.g. , journals) will blur in the future. Geneticist Dr. Kelly Frazer of Perlegen Sciences, Inc. discussed the importance of defining common human phenotypes via haplotype mapping efforts. She also noted the importance of interdisciplinary groups, but stressed that researchers must be encouraged to work together while keeping their specialist foci.

Dr. Andrew Fire , a developmental biologist at the Carnegie Institute of Washington, presented as a key problem determining how cells recognize "foreigners," such as how cells use various nucleic acid structures ( e.g. , dsRNA) as red flags, as well as what roles small RNAs play during development. He echoed Dr. Frazer's view about retaining scientific specialization while broadly training the next generation of researchers, encouraging NIGMS to contemplate the necessary ingredients for first-rate graduate training programs in the future. Bioengineer Dr. Drew Endy of the Massachusetts Institute of Technology expressed his broad vision to learn the rules of composition and design of biological systems. His view is that re-factoring existing biological systems, such as bacteria, will point to areas in need of further study. Anesthesiologist Dr. Roderic G. Eckenhoff of the University of Pennsylvania Medical Center presented anesthetic drugs as "a scientist's biggest nightmare and a patient's best friend," since these drugs are given to 40 million people and scientists have little idea how they work. He noted the importance of a research focus on "low-affinity" pharmacology, since anesthetics are low-affinity drugs. Dr. Eckenhoff stressed that structural biological studies should focus on disordered regions of proteins, and that a better understanding of structural dynamics is required using techniques such as NMR spectroscopy. Dr. Frederick Cross , a yeast geneticist at the Rockefeller University, questioned how useful current model systems, such as yeast, will be in 2010, stating that current model systems should be further developed and exploited as tools to fully probe specific biological processes, such as the cell cycle. He described a cultural divide between biologists and quantitative scientists, stating that biologists are resistant to using computational and systems approaches, and many physicists view biology as "messy." Dr. Cross also spoke to the serious problem of the poor quality of quantitative training in undergraduate biology education.

Dr. J. Perren Cobb , a trauma surgeon at Washington University, described the critical care patient as a relatively new phenomenon, arising with the mid-1950s introduction of ventilators in intensive care units. He described that researchers know very little about this systemic human condition, to the extent that recovery from critical illness is an "unpredictable waiting game." Dr. Cobb encouraged NIGMS to support clinical research, since generalists can be viewed as especially important in medicine as "molecular connectivists," defining specific factors that contribute to illness or survival. Dr. Henry R. Bourne , a molecular pharmacologist at the University of California, San Francisco, predicted that systems biology will continue to be a central focus. He identified a tension between investigator-initiated research projects and large, collaborative research groups. The latter must not be allowed to become "behemoths that go on forever"; NIGMS must preserve the option to declare large experiments failures. Geneticist Dr. Ethan Bier of the University of California, San Diego forecasted that comparative genomics will be a valuable tool to understand human disease, suggesting that advances will emerge from analyzing a common problem using data from multiple different model organisms. He called for the development and use of controlled vocabularies and informatics approaches, and he also stressed the importance of fostering good education at all levels, including grade school. Dr. Thomas R. Cech , president of the Howard Hughes Medical Institute, gave praise for the Institute's long-standing support of basic research "where the sparks fly." He then suggested revisiting the R01 granting mechanism and modifying the current grant application requirement for preliminary data. Molecular biologist Dr. Richard A. Young wrapped up the introductory remarks section with some comments of his own. He predicted that much biological discovery in the future will occur through the production and analysis of high-throughput experimental data such as that afforded by DNA microarray technology. He spoke to the importance of interdisciplinary efforts and information fusion, e.g. , combining data from many different sources. Dr. Young pointed to a simulator of cell behavior as a key future tool, and he presented as a visionary topic the notion of probing genomic influences on behavior and on entire populations.

Discussion

A general discussion followed, in which participants discussed in more detail some of the ideas that had been raised individually. One topic that came up in the early discussion and recurred throughout the meeting was the "mathematization" of biology. Much discussion arose about whether tool and resource development should occur simultaneously as investigators pose questions and solve problems, or whether tool development should occur separately. Participants were in general agreement that a schism exists between life scientists and quantitative scientists, and this separation reveals itself in several ways. The first is the misconception that engineers, mathematicians, and physicists should "serve" biology. Several participants argued that there is an inherent difference between the approaches to finding and solving problems used by these two groups and that the relationship should rather be one of cooperation. However, limiting a potential synergy is the generalized notion that biologists have not successfully communicated the most pressing biological problems to quantitative scientists. Following on this theme, the quantitative components of undergraduate and graduate training in biology is currently severely inadequate.

The early discussion session foreshadowed another focus common to this meeting and other previously held advisory meetings: infrastructure. Statements about practical deterrents to accomplishing research ranged from institutional roadblocks (teaching responsibilities, writing grants) to complaints about the peer review process (resistance to funding risky projects and applications describing proposed collaborations among new investigators). Discussion in this vein also visited the topics of training (of students and physician scientists), methods development, and the high cost of large capital investments. Suggestions to overcome some of these problems included relieving recently trained physicians' debt load; encouraging research environments that are more "corporate," competitive, and goal-oriented; and miniaturizing equipment (and therefore costs). Other suggestions included funding travel grants (to learn new techniques in off-site labs), or the opposite: "on-site" sabbaticals, in which investigators are forced to stay at their home institutions, declining all meeting and seminar travel requests, so researchers have the time and focus to collaborate effectively with at-institution colleagues. Some participants thought that insuring access to capital-intensive technology via regional centers across the country was an important option to consider. Regarding the review process, suggested strategies included requiring or rewarding collaborative, interdisciplinary research endeavors, as well as possibly encouraging academic/industry partnerships.

Breakout Group Summaries

Participants were assigned places in one of three (color-coded) breakout groups, charged with considering the following questions and reporting back to the group discussion and conclusions.

What are the most exciting research questions that you (and your colleagues) expect to address in 2010?

  • What experimental approaches would you like to use?
  • What tools or resources will be needed?
  • What kinds of collaborations do you anticipate initiating?
  • What emerging combinations of science will you (and your colleagues) be likely to pursue?

Green Breakout Group

The green breakout group members included Dr. Mary Lidstrom (chair), Dr. Ethan Bier, Dr. Perren Cobb, Dr. Fred Cross, Dr. Mark Gerstein, Dr. Judith Klinman, Dr. Alejandro Sánchez Alvarado, and Dr. Stephen Sligar. The group's discussion revolved around five major scientific themes: heterogeneity of populations, dynamics, de novo design, integration, and relationship of function to phenotype. A recurring topic underlying future advancement in many of these areas is the requirement for measuring gene expression with increased spatial and temporal resolution. Also important will be collaboration between biologists and investigators from a variety of disciplines including analytical chemistry, clinical science, computer science, engineering, materials science, physics, and theory.

Heterogeneity of Populations
The group agreed that future scientific progress will hinge on understanding biologically relevant units rather than population averages. Current microarray data can contain many false positives, making the predictability and utility of these data unreliable. New mechanisms will be uncovered based on heterogeneity; many of these mechanisms have not been identified due to the common convention of collecting data on poorly defined, averaged populations ( e.g. , of organisms, RNA, proteins, other molecules). Suggested approaches to gather information on a homogeneous set of organisms or molecules of interest include the use of real time, in vivo measurements, and single-cell proteomics. Tools that will enable these endeavors include modernized imaging resources, such as new optical techniques and dyes, as well as expanded use and applicability of quantum dots (semiconductor nanocrystals that exhibit unique properties different from the bulk materials, behaving as quantum entities). Many of these tools will be very expensive. Other new tools that have the potential to greatly enhance single-molecule detection include new molecular sensors and nanorobots. Miniaturized, disposable systems may facilitate research efforts in this area by reducing experimental costs, and automated, interconnected data acquisition will expedite the transfer of data and knowledge to the broader scientific community.

Dynamics
Research on how systems move and change can best be advanced by improving current tools, such as spectroscopic methods ( e.g. , NMR, mass spectroscopy, laser techniques). Advanced imaging and sensing methods will be required to measure metabolite levels in cells and organisms, and to document fluxes and metabolic rate constants of small molecules and gases. Increasingly sophisticated computational methods may help introduce predictive models and render theoretical models into practical use. The development of systems databases will facilitate data mining and searching for biologically meaningful patterns. These investigations may help to explain the phenomenon of "catalytic excellence"--why enzymes are so much more effective than can be accounted for based on current data on binding, orbital steering, etc . A better understanding of the dynamics of molecular and cellular processes will be important to explain how organisms tolerate fluctuations in environment. For example, understanding how an organism "breaks down" ( e.g. , the death of a critical care patient) may be approached through an analysis of how systems fail (after death, many cells remain alive for some time; however, the vital connections have been broken). Resources to enhance the development and widespread availability of these tools include facilities for advanced spectroscopy and imaging, as well as accessible computational packages and public databases.

De Novo Design/Engineering
The group discussed likely future gains from re-creating molecular systems and assemblies from scratch. Organic chemistry has traditionally worked this way; observations are later confirmed through total synthesis. Experimental approaches to define and build genetic circuits may improve mechanistic understanding; dysfunction of such circuitry will also be instructive. Standardized computational packages and tunable eukaryotic expression systems should lead to the development of predictive models and perhaps ultimately, to a "virtual cell." Knowledge of human physiology may be advanced through standardized collection of data from patients in intensive care units. Over time, these data could be mined for developing standard operating procedures and treatments for critically ill patients. Future efforts along these lines may someday lead to a "virtual human" or a "virtual patient," and/or help lead to the development of algorithms to define processes such as regeneration and wound healing.

Integration
Integrating data with multiple levels of biocomplexity will be essential to moving forward in years ahead to develop a mechanistic understanding of life processes. The group discussed the importance of acquiring and analyzing experimental data at several levels with many different organisms. This integration can best be approached using computational and bioinformatics tools, such as a "computational translator," that can visualize, evaluate, extract and manipulate data and biologically significant patterns across disparate species. Tools and resources that will make these efforts possible include the development and use of controlled vocabularies and the development of ways to standardize publication and data deposition into public databases. Some participants noted that drawing statistically meaningful analysis from large data sets is very difficult, but that other disciplines, such as economics and the intelligence community, have succeeded in similar pursuits.

Relationship of Function to Phenotype
Attempts to define phenotypic determinants will be an important step toward understanding the mechanistic underpinnings for development and disease. Global approaches to measuring gene expression with increased spatial and temporal resolution will guide efforts to translate genetic information into functionally meaningful knowledge, and to understand how multiple gene defects contribute to the onset and maintenance of a diseased state. Establishing comprehensive mutant and misexpression collections, perhaps through national or regional centers of excellence, could facilitate this process. The group agreed that researchers should focus on complex stimuli and subtle phenotypes, not just on "severe" mutations that produce extreme phenotypes. Several participants agreed that mutants should not be the focus of study; rather, mutant genes should only be used to search for the "normal" gene. Tools and resources that will lead to advances in defining function and phenotype may include: systematized protein expression, microsystems for phenotypic analysis, as well as the cross-species investigation of specific developmental issues/diseases. One current example of such an approach is the "Homophila" project, in which researchers have scoured the Drosophila genome for orthologs of human disease genes.

Blue Breakout Group

The blue breakout group members included Dr. Richard Morimoto (chair), Dr. Henry Bourne, Dr. Drew Endy, Dr. Andrew Fire, Dr. Stephen Harrison, Dr. Sydney Kustu, Dr. Jean Schwarzbauer, and Dr. Lukas Tamm. The discussion from this group culminated in identifying approaches and resources to enable scientists to model cell behavior, experimentally and/or computationally. The group described key research topics and necessary tools and resources that bear on this long-term goal.

Research Questions

Small Molecules and Large Protein Complexes
Building on current knowledge of cell composition, the next steps in understanding cell behavior will hinge on identifying and analyzing large protein assemblies. Key to achieving this goal will be the ability to visualize in real time large complexes (protein-protein, protein-membrane), especially ligand-dependent interactions. In a related vein, the group highlighted the importance of encouraging progress in developing new technologies for detecting single molecules. Technological breakthroughs will likely arise from advances in light microscopy and new chemistry. Watching molecules conduct enzymatic reactions and communicate messages throughout the cell will require new imaging software that is accessible and a new generation of very sensitive molecular probes (akin to the current probes used for the imaging methods FRET, FRAP, and FLIP). Participants also called for an increased focus on measuring metabolic concentrations and metabolic flux, since little experimental data has been produced to quantify metabolic flux inside various cell types. Approaches using small molecules, chemical biology, and combinatorial chemistry may advance progress in this area. While participants highlighted the importance of generating new tools, group members agreed that discoveries will still rely on individual "aha" moments in individual human brains, and that an era of science is needed in which technology and knowledge gathering go hand in hand.

Modeling
Many unanswered questions about intracellular molecular connectivity and cell-cell communication must be tackled before investigators can attempt to model cell behavior. There are currently many basic problems in biology that cannot be modeled, and even for systems that have a more complete set of experimental data, there is some disconnect between the timeframe required for biocomputation and new data acquisition. Creating mini-networks connecting multiple cell types in three dimensions ( e.g. , reconstructed "organs") may be a useful intermediate step toward building cell and organismal models. Currently, carbohydrates are a bottleneck in understanding the three-dimensional structure of tissues. Synthetics could serve as interfaces between cells, in tissue reconstruction efforts. To accomplish these goals, life scientists will require expertise from varied fields, especially engineering and materials science. Other useful strategies worthy of pursuit along these lines include targeted, genome-wide mutagenesis and exploitation of stem cell model systems.

Tools/Resources
The group identified three broad areas in which increased support may yield advances: infrastructure, education, and modifications to the grant review process.

Centers and Training
Technology development grants may be a vehicle to foster the development of tools to help solve specific research questions. Partnering with industry and the region may help finance some of these efforts. Regional or national centers could be established to catalyze research, training, and the development and dissemination of tools for specified scientific areas of study. One example provided was the Advanced Photon Source (APS) at Argonne National Laboratory, which hosts an annual workshop to educate the scientific community about topical issues in structural biology and APS resources. Other examples include the Stanford Microarray Core Facility, HHMI's Janelia Farm, and workshops and courses offered by organizations such as Cold Spring Harbor Laboratory. Expanded use and modernization of existing tools (proteomics) and approaches (chemical biology) are likely to facilitate research advances. Some members of the group stated that new "user unfriendly" software is needed, to enable researchers to ask unique questions while collecting and analyzing data. Group members agreed that a severe training problem is manifest in biomedical research, in which few graduate students and postdoctoral fellows receive adequate quantitative training. One vehicle for change might be to draw physics students "back to biology." Training grants can be used as tools for change, and the group recommended upgrading training programs to encourage the most innovative science by renewing emphasis in certain areas as well as by establishing new areas of emphasis. In creating new training grants, administrators and principal investigators should carefully define educational goals beyond degree acquisition, and realistically consider how long it takes to advance a student from college to becoming an independent investigator. Several group members stated that trainees did not have enough time to acquire the necessary skills to conduct interdisciplinary research. Access to training programs should be broadened to diversify the biomedical research work force. The group recommended that clinical scientist training programs should be maintained, and that more flexibility should be encouraged to permit M.D.'s to conduct research in no-traditional settings ( e.g. , the Carnegie Institution). Some, but not all, members suggested supporting translational research. While urging continued, stable support for individual investigators, the group also recommended expanded use of NIGMS' "Collaborative Project Grants," both within and across academic institutions. Some, but not all, participants suggested that established, senior scientists ( e.g., MERIT awardees) could be targeted to participate in the peer review process.

Red Breakout Group

The red breakout group members included Dr. Margaret Riley (chair), Dr. Thomas Cech, Dr. Roderic Eckenhoff, Dr. Kelly Frazer, Dr. Erich Jarvis, Dr. Richard Lerner, Dr. Ed Scolnick, Dr. George Whitesides, and Dr. Richard Young. The group's discussion centered on finding ways to answer the fundamental question: "What constitutes life?" Discussion along more practical lines centered on how to develop a better understanding of the "basic unit of life," the cell, and focused on the following themes: functional genomics, complexity, mechanisms of function/dysfunction, and cell regulation/cell interactions. Two separate but critical themes were tools, which the group described as "opening the window of biology," and collaboration/education.

Functional Genomics
The group discussed whether a point could be reached in which genomics would be equivalent to functional genomics, described by one participant as "using genomics to figure out physiology faster." One strategy useful for getting to that point will be to exploit the use of genomic information to rethink current views about genome function. Researchers must gain a better understanding of the role played by non-coding, "junk" DNA. The tools of proteomics may offer insight into "the protein folding problem" and help to elaborate on protein function and dynamics.

Complexity
Understanding the complexity of how, when, and where cells interact will be critical information for understanding the ingredients of health and disease. One participant described the usefulness of having an Internet-based "wiring diagram"--the ability to open a Web browser to view networks connecting all metabolic pathways in a cell. Efforts to develop signaling pathway networks may point to mechanisms of disease like cancer, which can be viewed as a "collaboration" between founder cells and epithelial cells. Nanobiological approaches may help to elucidate cellular dynamics. The group posed several questions: "Can a cell or its components be re-programmed?" Can the emergence of novel attributes during evolution or during development be predicted? What are the biological determinants for morphology? How does the brain generate language? One participant noted that evolution-based reasoning is a central part of thinking about biological problems. The use of non-traditional model organisms may help to address some of these issues, along with comparative genomics approaches.

Mechanisms of Function/Dysfunction
An important theme identified by the group was the relevance of understanding mechanisms of disease. Future progress may result from determining to what extent errant gene expression impacts disease, and whether there are diseases that result from faulty three-dimensional arrangements inside and/or between cells or from problems with the fluidity of membranes. A better understanding of cellular dysfunction will likely improve pre-symptomatic and molecular diagnosis. Another application of this knowledge may include the ability to accurately diagnose the source of an infection, predict its course, and tailor ways to treat the infection. More emphasis on understanding genetic flow across species will help clarify mechanisms of infectious disease and maintenance of natural microbial reservoirs. The group felt that enabling clinical investigation, which participants described as a critical but dying science, was important and recommended that NIGMS assemble a group of physicians to offer advice on successful strategies to achieve this goal. Problems facing clinical research include large debt load incurred by physicians and traditional medical school training constraints. Some participants reported that they knew few physicians who had been well-trained at the molecular level, and that laboratory rotations are insufficient to train physicians how to conduct biomedical research.

Cell Regulation/Cell Interactions
Understanding cellular and genetic networks may lead to developing "rules of design" for building cells, organisms, and signaling networks. Central to progress in this area is investigating interactions between cells, understanding differentiation processes, and more thoroughly characterizing metabolism dynamically. Advances in single-molecule detection and the ability to detect and track large molecular complexes are both important for progress in this area. Advances in imaging will be expensive but vital. The group discussed the importance of mapping genotype to phenotype on a larger scale than the current single-molecule approaches. Identifying relationships between genetic variation ( e.g. , SNPs, haplotypes) and phenotype will continue to be important areas of study. Some participants noted that very few microbial physiologists exist, despite the fact that training in this area will be an important component of future biomedical research, especially for biosafety efforts. The group identified a need for developing non-linear descriptions, and members questioned whether a cell behavior simulator could ever become a reality. More circumscribed modeling efforts, such as computational simulations of processes such as anesthesia, may help drug discovery efforts for anesthetics and other types of "low-affinity" drugs.

Tools
The need for new tools was a pervasive discussion topic, and all participants agreed that real progress in tool development is likely to be very expensive. Discussion in this arena revolved around the issue of tool inventors vs. tool users, and participants disagreed to some extent about the most fruitful avenues for the development of useful tools to address pressing scientific problems. Some of the types of new tools needed are tools to monitor insoluble proteins; miniaturized devices; and physics-based tools, such as microcalorimeters. Bioassays are needed to analyze organellar behavior and basic cellular processes such as cell motility and mitotic spindle function. One participant described his experience with the Defense Advanced Research Projects Administration (DARPA). This federal component works by bringing together a large group (~100) of researchers to work intensively on a particular area of scientific need/interest decided in advance. DARPA-funded projects have heavy program management, a defined (non-renewable) lifetime of 3-4 years, and a large budget ($5-10 million/project). Discussion centered on the merits of a DARPA-like approach, in which the development of a specified application is driven by an outcome pre-defined by a program announcement. New methods for investigating the toxicity of potential drugs may decrease the number of steps in the drug discovery to drug testing process. One participant described toxicology as the "single biggest rate-limiting step" in drug development. Questions were raised about whether NIH could assign a more primary focus to tool development, particularly to the creation of centralized, universal, integrated databases with new query methods.

Collaboration/Education
Information sharing should be encouraged if not required; funding should be contingent on depositing data into the public domain. New funding mechanisms to assemble "critical masses" of scientists working on a common scientific problem may speed the pace of discovery. DARPA-like approaches may also help to encourage collaboration. Forced communication at least twice a year (conferences among the participant researchers), along with the requirement for developing both a product and a scientific community, has made this a successful model ( e.g. , the Internet was created through a DARPA-sponsored effort). The group recommended efforts to encourage collaboration and integration at all levels, but stressed the importance of maintaining specialists in an emerging era of integration. Group members nearly unanimously agreed that trainees have very poor computational skills. One participant mentioned a successful course at his own institution in which computational functional genomics is team-taught by a biologist, a mathematician, and a computer scientist. Course goals include creating a dictionary of terms and teaching students to code questions via algorithms. New methods are needed to teach innovative thinking and to encourage risk taking.

Post-Breakout Group Discussion

Following the breakout group meetings, general discussion of the entire group visited themes common to the discussion held in individual sessions. Dr. Richard Young invited the group once again to voice their individual scientific visions, and he began the discussion by asking participants to also consider the future possibility of creating a "virtual human." The suggestion prompted widely variable responses among discussants. Most participants viewed the goal of creating a virtual human untenable in the foreseeable future of biomedical research, particularly if the goal is to create a predictive model of human physiology. Participants also viewed the goal as something that should be approached by the entire NIH, as a "centralizing force" relating work in many disciplines to human health. The discussion touched on the possibility of creating other virtual entities, such as a virtual cell, tissue, appendage, or disease state ( e.g. , sepsis), then migrated to a more general dialogue about modeling.

Building models is a complicated endeavor, and many biological processes cannot yet be modeled due to a generalized lack of information about individual system components. To that end, Dr. George Whitesides noted that at present a virtual cell could not work as a simulator because, relatively speaking, few of the pieces of the cell (as a system) are known. Another problem is finding researchers equipped to transform "messy" biological data into useful algorithms. To catalyze this process, a fusion of biology with the physical and engineering sciences is needed, and Dr. Whitesides suggested that NIGMS could play an important role in establishing connections between these fields. Dr. Mark Gerstein suggested an alternative approach of creating computational resources that use sophisticated statistical methods to extract meaning from existing data. Dr. Drew Endy, a modeling expert and engineer, offered commentary on requirements for building models and current gaps in knowledge of cell biology that pose serious limitations for modeling cell behavior. Cross-scale modeling, i.e. , between cells, organs, and organisms, is difficult now. Dr. Endy stressed the importance of defining the endpoint of modeling efforts, citing the surrealistic painting "Clairvoyance" by the artist René Magritte, in which a man stares at an egg while painting the image of a bird. Will models be used to explain or predict biological behavior? He stated that current issues that prevent the development of realistic human biology models include: an incomplete understanding of the physics of human biology, poorly defined protein interactions, rate equations that are not experimentally grounded, and an understatement of the role of environment in biological determinism. Dr. Endy cited the primary reasons for model failure as: an incomplete understanding of the parts, computations that are not well-grounded, ill-posed perturbations ( e.g. , data such as incomplete inhibition of kinase activity), misinterpretation of results ( e.g. , measuring populations vs. single cells), and artificial boundary conditions. Dr. Endy questioned whether modules really exist in biology, and spoke to the difficulty in creating models consisting of non-linear components. He stated that his own goal for the coming decade is to rebuild the human genome and for that effort the following information is still needed: a more extensive and complete "parts list," a thorough understanding of the relevant chemistry, specification of system function, and a consideration of environmental ( e.g. , non-genetic) effectors.

The discussion continued, reiterating points previously made about the importance of tool development and the current "crisis" in the quantitative training of biology students. Following up on Dr. Young's call for more visions, participants provided additional thoughts on the future of biomedical research. Dr. Whitesides suggested considering engineering outcomes other than the production of new drugs, such as toxicity screening tools and pre-screening of patients for clinical trials, and building interfaces between microelectronic devices and living cells, to collect and code data. Discussion of the power of comparative genomics led to suggestions such as interspecies conversion (Dr. Ethan Bier), trait acquisition (teaching pigeons to sing, Dr. Erich Jarvis), identifying emergent infectious agents (Dr. Margaret Riley), and understanding genome organization and the differentiation potential of genomes (Dr. Alejandro Sánchez Alvarado). Dr. Andrew Fire stated that an important goal will be to understand all of the genetic and epigenetic mechanisms necessary to create a stable change in development or disease. A deeper understanding of cellular and multicellular networks may facilitate testing of certain classes of drugs, such as anesthetics (Dr. Rod Eckenhoff). Efforts to model human responses to infections and inflammation may deepen understanding of human physiology (Dr. Perren Cobb). Dr. Richard Morimoto described a need to identify and understand the environmental cues that perturb cell function, as well as to investigate further how cells react to stress using molecular chaperones. Dr. Frederick Cross described the value of having an experimentally validated, predictive model of the cell cycle. Dr. Ethan Bier suggested a focus on the genome's "regulatory code," noting the importance of analyzing promoters, introns, and other non-coding genomic regions. Dr. Young projected gains from solving fundamental molecular mechanisms, such as nuclear-cytoplasmic transport, DNA replication, transcription, cell division, molecular chaperone function, and cell-cell communication. A pervasive recommendation among participants was the need to create and maintain a well-trained research work force.

Roster

Richard A. Young, Ph.D. (chair)
The Whitehead Institute for Biomedical Research

Ethan Bier, Ph.D.
Division of Biology
University of California, San Diego

Henry R. Bourne, M.D.
Department of Cellular and Molecular Pharmacology
University of California, San Francisco

Thomas R. Cech, Ph.D.
Howard Hughes Medical Institute

J. Perren Cobb, M.D.
Cellular Injury and Adaptation Laboratory
Washington University in St. Louis

Frederick Cross, Ph.D.
Laboratory of Yeast Molecular Genetics
Rockefeller University

Roderic G. Eckenhoff, M.D.
Department of Anesthesiology
University of Pennsylvania Medical Center

Drew Endy, Ph.D.
Division of Biological Engineering and Department of Biology
Massachusetts Institute of Technology

Andrew Fire, Ph.D.
Carnegie Institution of Washington

Kelly Frazer, Ph.D.
Perlegen Sciences, Inc.

Mark Gerstein, Ph.D.
Molecular Biophysics and Biochemistry Department
Yale University

Stephen C. Harrison, Ph.D.
Department of Biological Chemistry and Molecular Pharmacology
Harvard Medical School

Erich Jarvis, Ph.D.
Department of Neurobiology
Duke University Medical Center

Judith P. Klinman, Ph.D.
Department of Chemistry
University of California, Berkeley

Sydney G. Kustu, Ph.D.
Department of Plant and Microbial Biology
University of California, Berkeley

Richard A. Lerner, M.D.
The Scripps Research Institute

Mary Lidstrom, Ph.D.
Department of Chemical Engineering
Department of Microbiology
University of Washington

Richard Morimoto, Ph.D.
Department of Biochemistry, Molecular Biology, and Cell Biology
Rice Institute for Biomedical Research
Northwestern University

Margaret Riley, Ph.D.
Department of Ecology and Evolutionary Biology
Osborn Memorial Labs
Yale University

Alejandro Sánchez Alvarado, Ph.D.
Department of Neurobiology and Anatomy
University of Utah School of Medicine

Jean E. Schwarzbauer, Ph.D.
Department of Molecular Biology
Princeton University

Edward M. Scolnick, M.D.
Merck Research Laboratories

Stephen G. Sligar, Ph.D.
Departments of Biochemistry, Chemistry, and the College of Medicine
University of Illinois

Lukas K. Tamm, Ph.D.
Department of Molecular Physiology and Biological Physics
University of Virginia

George M. Whitesides, Ph.D.
Department of Chemistry
Harvard University