Quantitative & Systems Pharmacology Workshop

September 25-26, 2008

Quantitative and Systems Pharmacology Workshop Executive Summary
Group 1: Horizontal System Integration in Pharmacology
Group 2: Vertical Integration in Pharmacology
Group 3: Quantitative Biology and Pharmacology
Group 4: Education Training
Group 5: Data Management - Overall Impressions
Addendum: Planning Document


The Quantitative and Systems Pharmacology Workshop held on September 25-26, 2008 sought to address the question of where systems biology, modeling, and more quantitative measurements could be applied to pharmacology and drug discovery/action now and in the foreseeable future. Meeting goals included highlighting where the state-of-the-art resides in relevant disciplines today, fostering integration of research efforts in these areas with others, identifying barriers (problems merging two disciplines), increasing collaboration between pharmacologists, clinicians, and systems biologists, and establishing pressing and long-term research needs (training, instrumentation, algorithms, etc) for advancement of the ability of systems biology to inform drug discovery and drug action. The meeting report that follows includes an executive summary of the workshop authored by Douglas Lauffenburger, who along with William Jusko, co-chaired the workshop. The executive summary is followed by reports from each of the five-breakout session co-authored by the discussion leader and recorder for each session. (Group I: Michael Phelps and Reka Albert; Group II: David D'Argenio and Rai Winslow; Group III: Juan Lertora and Peter Sorger; Group IV: Henrik Dohlman and Fei Hua; Group V: Yoram Vodovotz and Avi Maayan) Finally, also included is a copy of the pre-meeting planning document that assembled and recorded community input on the areas covered by the workshop; this was authored by NIH staffers, Peter Lyster and Sarah Dunsmore.

NIH staff have begun planning for a follow up workshop tentatively titled Quantitative and Systems Pharmacology II.

Quantitative and Systems Pharmacology Workshop Executive Summary

D.A. Lauffenburger, Massachusetts Institute of Technology

The Quantitative and Systems Pharmacology Workshop, held 25-26 September 2008 at the NIH campus, aimed to provide state-of-the-art-knowledge and perspectives on the interface of systems biology and pharmacology to a highly diverse spectrum of researchers in academia, industry, and government. Addressing the question of where systems biology, modeling, and more quantitative measurements can be applied to pharmacology and drug discovery/action now and in the foreseeable future was the central focus of talks and discussion sessions. Five major aspects of the systems biology/pharmacology interface were explicitly featured: [1] horizontal systems integration; [2] vertical systems integration; [3] quantitative biology and pharmacology; [4] data management; [5] education and training.

Widespread appreciation arose for the need to create mechanisms for bringing the systems biology and pharmacology communities together more frequently and deeply, because of the clear promise for common benefit. The complexity inherent in understanding and predicting the pathophysiology and drug effects in patients ought to be gainfully attacked by including the methods and concepts being developed by systems biology investigators and, in turn, the envisioned impact of systems biology ought to be pursued vigorously in this important realm of molecular medicine. The arena of systems biology would benefit from greater appreciation of the diverse complexities inherent in the actions of many drugs. Systems and homeostatic perturbations induced by drugs can serve as probes of the validity of systems models.

One dimension of the scientific challenge is the need for more comprehensive 'horizontal integration'. That is, even at the level of cellular pharmacodynamics, understanding drug action requires integration of effects within the myriad interconnecting multi-pathway networks of signaling, metabolism, and gene expression; moreover, operation of these pathways involves molecular communication across cell boundaries into the microenvironment. Hence, quantitative measurement and modeling must emphasize incorporation of multi-pathway information and in tissue environmental context, which will require major advances mainly in experimental technologies including tissue-level imaging with molecular resolution. The second dimension of scientific challenge naturally follows, the need for more explicit 'vertical integration', in which "bottom-up" models rooted in detailed molecular mechanisms at the cellular level must meet "top-down" models describing organ- and organism-level physiology with the objective of the higher-level observations be interpretable and predictable more explicitly in terms of measurable molecular and cellular properties. This will require advances mainly in computational methodologies via which consequences of multi-pathway molecular mechanistic detail at the lower level can be propagated efficiently as the scope escalates up from cells to tissues to organs to organisms. A high priority goal is to leverage the systems approach to shift the basis for biomarkers from correlative to mechanistic, and likely multi-variate, linking quantitative biology to quantitative pharmacology. Because of practical constraints on proving and achieving the aspired combination of horizontal and vertical integration in human patient studies, one approach proffered is dedicated full-scale tests in animal studies. Although successful accomplishment of an envisioned quantitative systems pharmacology paradigm therein would not have short-term impact in the clinic, it would give confidence in the effective utility of the new paradigm. A third, complementary challenge arises from the problem of managing—i.e., organizing, storing, accessing, and visualizing—the various kinds of experimental data in concert with the diverse classes of computational models that attempt to capture their salient features. This problem is dissimilar to the more straightforward issue previously experienced by the genomics field, in which data can be relatively easily managed due to its fairly homogenous structure. The immense degree of heterogeneity of horizontally- and vertically-integrated data for drug actions within complex molecular networks for understanding and prediction of organ- and organism-level physiological behavior will require coupled data- and modeling-management approaches that cannot currently be found "on the shelf". Finally, there was essentially unanimous recognition that a new cohort of scientists and engineers will need to be educated at the systems biology / pharmacology interface, as the current population overlap is vanishingly small. Consensus landed on postdoctoral training as the most effective locus for dedicated educational efforts, because of the need to have a strong research capability foundation first before tackling the daunting inter-disciplinarities involved. One unusual avenue gained significant favor, that of multi-institutional postdoctoral training programs, since few institutions possess the necessary expertise across the many contributing fields.

Group 1: Horizontal System Integration in Pharmacology

As a working definition of Horizontal Systems Integration we refer to the integration of multiple interconnected events at the molecular mechanism to cellular outcome (phenotype) level of pharmacology. The system may involve, for example, multiple signaling or metabolic pathways and the interconnections of signaling to metabolic pathways; multiple normal & disease cell types; communication between cells to execute multi-cellular functions; and signaling, metabolism and cellular response to drug treatments to define therapeutic effectiveness, as well as compensatory and refractory responses. The phenotype and biological activity of these systems is the target of pharmacological intervention, and is the target of in vitro and in vivo (imaging) molecular diagnostics to read out the differential states of normal vs diseased cells.

Progress towards empowering systems approaches must be based on integrating experimental, modeling, and theoretical approaches. Quantitative and kinetic measurements are critical for understanding the biochemical foundation of these events and identifying molecular targets with a therapeutic index. Emphasis must be placed on developing new measurement technologies for system-wide measurements, and new theoretically founded algorithms for interpreting the resulting data and building a more complete description of the systems. Technology and methodology advances can produce powers of 10 in accelerating and expanding approaches to solving this systems problem. Novel technology platforms will be enabling in creating the measurement sciences necessary to achieve these goals.

The new measurement platforms need to be evolvable and scalable to meet the growing and changing needs of experimentation and implementation of great ideas by great scientists, similar to the revolutionary outcomes resulting from Moore's Law in integrated circuits. Useful characteristics of such technology platforms will include real time measurements, measurements of metabolic reaction rates and fluxes through pathways, and the development of new classes of protein capture agents for moving proteomics techniques towards the scale of technology platforms developed for genomics. The theory, models and algorithms should aim to make system-wide descriptions and results more comprehendible and useful to biologists and clinicians.

This is further detailed in the outline below:

  1. Definition: Integration of multiple pathways & multiple normal & disease cell types in response to drug treatments
  2. Horizontal Integration:
    • from molecular level to cellular level
    • from mechanism to outcome (phenotype)
    • taking into account the microenvironment of the cell (cell to cell communication)
    • taking into account the context of the cell participating in organized multi-cellular functions (e.g., a tumor functions as a pseudo organ system)
    • relationship of in vitro cellular functions to those of cells in patients (molecular imaging)
    • toward full characterization of spatio-temporal physiology (activity) of normal and diseased cells
    • establishing the relationship between the profile & concentrations of disease based proteins in tissue to the profile & concentration of excreted proteins in plasma—plasma based molecular diagnostic
  3. Critical aspects: integration of experimentation, modeling and theory
  4. Points of emphasis:
    • quantitative measurements
    • kinetics of competitive processes & biochemical reactions
    • molecular targets
      • with a therapeutic index
    • drug development and testing
    • stratifying patients in terms of developmental stage of disease & predicted drug response in individual patients
  5. Areas that will enable quantitative systems pharmacology:
    • new measurement platforms (new technologies):
      • evolvable and scalable (Moore's Curve of platform technologies for biology)
        • e.g. integrated microfluidics
      • molecular diagnostics for solving research problems—run diagnostics on cells, tissue & plasma to understand disease transformations and from this to generate approaches for molecular diagnostics on patients
        • real time measurements, dynamic resolution, appropriate time sampling
      • tissue samples, reliable sample collection aligned to new experimental paradigms
      • issues in measuring proteins, modifications, biochemical activity
        • capture agents (e.g., antibodies, aptamers, phage display, and new approaches to yield a class of protein capture agents with high affinity & specificity produced through a process that is scalable, reliable and low cost)
        • taking these protein measurements towards the scale of genomic and gene expression methods (will be useful even before hitting the full system-wide scale)
      • metabolic flux; add kinetics of biochemistry to the systems determinations
      • how to engage industry in these goals
    • new algorithms
      • network analysis
      • inference
      • model building
        • detailed versus abstracted (course-grained)
          • (also an issue for vertical integration)
        • models offering different vantage points (e.g. Bayesian vs. stochastic)
        • evolvable
        • scalable
          • e.g. a detailed fluid dynamics model that leads to an abstracted empirical heat transfer coefficient (a function that fits the data)
        • no monolithic models
      • integration of different types of measurements
      • computational limits
      • interpretable results for clinical use

Group 2: Vertical Integration in Pharmacology


A number of existing efforts, under the rubric of multiscale modeling, are of conceptual and specific relevance to the challenge of vertical integration in understanding drug action. The work being done by investigators under the current NIH Multiscale Modeling PAR should be closely examined and the successes and failures of these efforts should be analyzed to guide efforts in pharmacology (drug discovery and development). (Also, the efforts and reports of the Interagency Modeling and Analys?is Group (IMAG)" can provide important insight.)


Vertical Integration in Pharmacology: Synthesis of knowledge and understanding of drug action at the molecular, molecular complex, sub-cellular, cellular, multi-cell, tissue, organ, multi-organ systems, organism, and population levels.


The prevailing framework for multiscale integration (to the extent it is considered at all by biological scientists) relies on the traditional experimental approach of studying subsystems, which can involve isolating the subsystem from external factors (or attempting to keep them constant) and studying the behavior of the isolated subsystem. Multilevel behavior is then inferred by combining the subsystems. This is largely the approach contemplated and practiced by most involved in the multiscale modeling initiative. The nature of such an isolated subsystem experimental/conceptual approach may limit ones understanding of the overall biological system, especially when the goal is to understand drug action in disease.

An alternate approach to vertical understanding ("integration"), that aims to identify and elucidate the guiding principles of control and communication defining the behavior of an organism across scales is also needed.

Challenges Identified During the Breakout

  • The main challenges confronting multi-scale modeling are: a) identification of the biological systems (components) with which integration will be done; b) quantitative experimental characterization of these components; c) development of reproductive and predictive models of each component; d) determination of the appropriate degree of model complexity that can be included in the model integration; e) model validation; and f) application of mathematical and computational methods to achieve the integration
  • Vertical integration can mean both bottom-up, top-down, or middle-out. In either case, the challenge is to establish a roadmap that allows one to move up in scale and also back down in scale in a guided manner.
  • Multiscale models of drug action that include disease progression are required in drug discovery and development.
  • Identification of the appropriate ways to integrate across different levels of biological organization remains a challenge. As one example, the complexity of how to relate gene expression to protein expression will require understanding of a complex network of interacting parts.
  • Vertical integration, as with all modeling, requires that data used in the modeling be carefully annotated with metadata describing experimental protocols, assays, preparations, etc. This can be called "context". A challenge is developing ontologies, data models and data exchange formats to do this.
  • Challenge—building re-usable model components for either vertical or horizontal integration requires establishing and using standards, as well as best software engineering practices. "Industrial grade" software engineering talent is needed.
  • As in modeling at a single level of integration, it is important that vertically integrated models be tested extensively to determine which data they reproduce, and where they fail. Failures can guide design of experiments that will help fill in the black boxes.
  • For the pharmacology and biological sciences supporting drug discovery and development, a critical step on the vertical integration later is the in vitro/in vivo transition. New experimental and modeling approaches that provide for predictions and validation across this level are needed.
  • Need for case studies in different biological domains showing mathematical and algorithmic methods for vertical integration.
  • Development of high-level toolkits for studying drug action that support the re-use of model.
  • Exploit the teaching aspects of re-usable components and high-level toolkits.


  • Need multiscale "grand challenge" efforts for understanding drug action in disease (especially chronic diseases) for particular biomedical domains that involve all levels of vertical integration. Examples may include:
    • Type 2 diabetes and other metabolic diseases
    • Cardiovascular diseases
    • Autoimmune diseases

Group 3: Quantitative Biology and Pharmacology


An integrative discipline that interfaces between drug discovery and development and uses biological modeling approaches for hypothesis generation and testing, spanning chemical, biochemical, and physiological processes relevant to drug effects (both toxic and therapeutic) in healthy and diseased organisms.


A systems biology approach to the study of drug effects is predicated on the utility of models as effective means of summarizing broad sets of data that can then engage a diverse community of participants for the conceptual understanding of outcomes and further research focused on elucidating underlying mechanisms and identifying new therapeutic drug targets.

Modeling can integrate experimental data derived from in silico, in vitro, and in vivo studies in animals and humans. It can include computer assisted drug design based on known or modeled drug-target interactions, homeostatic control systems that modulate drug effects at the molecular, cellular, organ, and system levels, pharmacokinetic-pharmacodynamic modeling, disease progression modeling, animal models of disease, and clinical trial simulations. Ultimately, rigorously validated models should be integrative, quantitative, and predictive. They should also account for individual variability in drug response due to environmental and/or genetic influences, including the placebo effect.

A systems biology approach should also result in discovery of novel biomarkers and a shift from correlative to mechanistic biomarkers relevant to the natural history of disease and to therapeutic and toxic drug effects.


  • To identify subsets of patients that respond, and are likely to benefit from a given pharmacological therapy.
  • To identify subsets of patients that may be at increased risk of serious adverse drug reactions.
  • To identify novel drug targets and novel biomarkers of disease pathophysiology and drug effects.
  • To understand the quantitative aspects of selectivity of drug effects and the integrated dose response surface for multiple effects elicited by the same drug in whole organisms.
  • To "reposition" established drugs that are shown to impact on novel biomarkers of disease, disease progression, and therapeutic and toxic drug effects.


  • Promote dialogue among diverse scientific disciplines that could contribute to develop integrated systems biology approaches applied to pharmacology.
  • Assemble multi-investigator teams spanning clinical and pre-clinical disciplines such that all the necessary skills are utilized in drug discovery and development.
  • Establish education and training pathways for quantitative and systems pharmacology.
  • Design a demonstration project highlighting gaps in knowledge and the path from animal to human pharmacology and therapeutics.


  • Establishing links among ontologies and modeling languages that will be required and need integration.
  • Access to data sets at all levels of biological organization both in the public and private domains with regard to drug effects and disease pathophysiology.
  • Need for computational resources and the informatics capability to establish a network.
  • Need for experts in translational therapeutics and clinical pharmacology (declining numbers).

Exemplars and Candidates for Demonstration Projects

  • Diabetes: Biomarkers of insulin resistance—Modulation by therapeutic drugs
  • Obesity: Biomarkers of inflammation—Modulation by therapeutic drugs
  • Pharmacogenetics:
    • Warfarin (thromboembolic diseases, thrombophilias)
    • Tamoxifen (oncology), clopidogrel (coronary artery disease)
    • Clopidogrel (coronary artery disease)
    • Carbamazepine toxicity (Stevens Johnson Syndrome)
    • Abacavir hypersensitivity (HIV-AIDS)

Group 4: Education Training

Definition and Challenges

These definitions and challenges were read but not discussed.


Defining the new combinations of competencies, masteries, habits of mind and skills required by the future investigators who will successfully use systems approaches to understand diseases and facilitate drug discovery and development. Defining the pedagogies, curricula and activities that will constitute training for these new combinations.

Pedagogical Challenges

Teaching (medical, pharmacy, as well as graduate students) thinking about biological systems (where vertical and horizontal integration are not constant) as distinct from engineering systems theories (where organization is fixed; and where the theory does not map well to biology) is a major challenge. The following are several reasons.

  1. the novelty of the field; concepts and ideas are still emerging, and those ideas are necessarily very different from the hyper-reductionist thinking that still dominates biomedical training
  2. its inter-, intra-, and multidisciplinary nature
  3. the diversity of its research community and thus the literature
  4. the lack of:
    • a common language, consequently the need (for individuals in an unorganized fashion) to borrow language from other fields (research in different domains have different meanings for the same terms)
    • well-defined subject boundaries (and the problems thereof)
    • generally accepted definitions; the absence of methods designed to achieve a long term vision
    • understanding and willingness in industry (and the FDA and NIH) to support such research
  5. Promote industry (and medical and pharmacy schools and the FDA) support such research or teaching. Collaboration is needed across the board to further develop this field. With the Critical Path initiative, the FDA is trying to promote more dynamic science-driven pre-clinical and clinical drug development paradigms. Their efforts in pharmacometrics have been significant as well.
  6. instruction in special courses and modules requires multi-domain backgrounds
  7. potential newcomers to the field (those initially attracted to the issues and problems) find it difficult to change their thinking
  8. the familiarity and consistency of a "domain-centered culture" is absent (it is multidisciplinary); there is no single theory (it is multiparadigm [multi-attribute; multi-perspective])
  9. both depth and breadth are pursued concurrently
  10. use of techniques to optimize therapy for patients, not just for research.

Discussion Points from Comments Received

The discussion points were introduced in the following order:

  1. What type of student should be recruited?
  2. What coursework should be required (quantitative sciences; laboratory sciences)
  3. How will programs develop a core of qualified faculty?
  4. How to prepare trainees for both academic and industrial jobs?
  5. How will programs be funded?
  6. What types of programs should be involved?
  7. Should a consortium approach be considered since data may reside with industry partners
  8. What type of training is needed (MS, PhD, Postdoctoral Fellowship)
  9. How much will the demands be from both academia and industry and government (e.g. FDA) in the near term and long term
  10. If so, what are the challenges to expand the program (e.g. funding, teachers, attract potential students etc.)

Discussion Summaries

The group recognized a significant need to expand current educational programs in quantitative pharmacology and pharmacometrics. The group identified several challenges for organizing such training programs, and proposed possible solutions.

Broader Training Considerations

As with any biomedical research enterprise, the ideal environment for training in systems pharmacology is likely to be a major research university with a comprehensive health affairs campus, with robust graduate and fellowship-training programs (including postdoctoral and clinical research fellows). Other institutions should not be excluded however. For example MIT lacks a medical school or hospital, but is a center of excellence in systems biology. Therefore, a collaborative training approach should be considered, such as a consortia mentioned below. There should also be a tradition of research excellence and a history of cooperation between the basic sciences (including pharmacology, physiology, biochemistry) and the quantitative sciences including bioinformatics, biostatistics, genomics, biophysics, engineering, computer-sciences.

Trainee Resources

Students learn much from one another, and should therefore be recruited from all disciplines and trained as a group. Thus students should be drawn from engineering, the quantitative sciences (physics, mathematics, statistics, computer sciences) as well as from the life sciences. In addition, much of the best work is being done in an industry setting. Thus trainees should also be recruited from the ranks of industry, as well as directly from colleges and universities.

Training Types

There should be a mix of masters, doctoral, and post-doctoral trainees (including clinicians). Post-doctoral trainees with experience in research, and looking to broaden their skills beyond the topic of their doctoral work, can help fill the short-term needs in the field. However, funding for additional postdoctoral training is also needed to encourage and help PhDs enter a new field. A pool of trainees with a Masters degree in systems pharmacology will serve the needs of some sectors in the short term. However students having only a Masters may in the long run have difficulty "keeping up" with the rapid pace of research. Another possibility is to recruit students that already have an MD or PhD, as is commonly done at schools of public health and epidemiology. Masters students pay tuition, and this could serve as a revenue stream that supports doctoral-level teaching and research.

University Consortia

Broad-based training in systems pharmacology will require a breadth of expertise that may be difficult to find on a single campus, at least for the foreseeable future. Thus training programs should consider organizing consortia that would share development of educational materials, provide teaching in cross-disciplinary topics, and promote collaborative and inter-disciplinary research programs. Consortia should also include industry, given that data and "real-world" experience may be concentrated in industry labs.

University-industry Cooperation

The needs are very much driven by industry, and in any case industry will need to participate in designing new curricula and training programs. There is a wealth of experience within industry, which may be drawn upon in filling the ranks of training faculty, for developing collaborations with academics, for hosting visiting scientists and interns, and as a source of students (such as mid-career scientists).

Distance & on-line learning

Some students currently in industry will not be in a position to relocate for training. Thus programs should consider developing on-line learning programs. Mechanisms for on-line learning will also facilitate the exchange of information among consortia members.


In addition to the traditional sources for research funding, there is a dire need for new NIH training grants dedicated to interdisciplinary research. Industry may be willing to provide financial support in some cases for cooperative research programs, and also to pay full tuition for their students. There is also a major need on the part of the FDA for individuals trained in pharmacometrics. Thus the FDA might be a funding source for new training programs.


Courses should include a mix of didactic lectures and hands-on practical (laboratory or computer) training. Here is a list of potential courses that might be offered in a systems pharmacology program:

  1. Modeling Core Coursework
    • PK/PD modeling
    • Physiological based/mechanisms based PK/PD modeling
    • Clinical trial design simulation
    • Disease process/progression modeling
  2. Pharmacology Core Coursework
    • Experimental and clinical pharmacology and toxicology
    • Pharmacogenomics
    • Drug absorption, metabolism, distribution and excretion
    • Biopharmaceutics
  3. Statistics Core Coursework
    • Applied Biostatistics
    • Advanced Statistical Methods
      • Regression Analysis
      • Multivariate Analysis
      • Longitudinal Data Analysis
      • Survival Analysis
    • Monte Carlo Methods
  4. Programming Core Coursework
    • Computational Methods/Application and Development
  5. Introduction to Statistical Programming

Group 5: Data Management - Overall Impressions

In a sense, the goal of the meeting was to bring old style modeling pharmacology efforts together with the "new" systems biology experimental and modeling advances. However, in order to be more innovative we need to integrate Systems Biology with PK/PD studies and other modeling efforts but it is impractical to be an expert in all those fields. Moreover, there are challenges specific to the integration of data in these two camps. We have diverse datasets and data types, i.e., data from clinical, pharmacological, cellular studies. How can we manage and integrate such data to advance translational research? A consensus of the group was that "there are so many databases... more databases than articles... by the time you learn what is in one database, another appears... it is impossible to be aware of what is available... a lot of efforts are being duplicated because of that." Indeed, there are also many modeling tools that do the same thing, being developed in parallel and without knowledge of the work of others. It was then agreed that this is also why we need more meta-data and exchange standards: This will improve searching, and speed up the process of understanding the content of databases.

One common goal could be to collect data for building unifying models. Alternatively, it may be argued that thinking about models is restrictive since sometimes viewing a collection of facts or exploring networks that display relations between entities can be also very informative. Models are only one way to extract knowledge from data. There are many different types of models at different layers. Hence, models should be linked such that the output from one model can be the input to another model. The consensus goal was then extended to a broader definition that states: our goal is to develop data management solutions allowing for data integration aimed at improved extraction of knowledge from horizontal and vertical studies.

How can we achieve this goal? The problem is that we are currently not efficient in storing, searching and reusing data and models. One solution is to add meta-data to existing and new data and models. Many models do not describe specifically the entities that are modeled, limiting their broad utility. This concept has been termed "harmonization", notably as relating to work carried out by the Alliance for Cell Signaling. This approach was attempted but achieved only partial success. It is also useful to utilize modeling software that forces authors to heavily annotate their data when submitting papers for publication. Useful tools are being developed to facilitate the conversion of models and data into sharable formats automatically: e.g. JigCell ability to export models in ODEs and SBML, and tools to convert protein interaction networks and signaling pathways saved in text files into BioPAX format. Scientists are motivated to publish papers and will be willing to dedicate their time and effort to provide meta-data along with their studies. It was then agreed that there should be established data reporting protocols.

In those settings in which there is extensive use of Laboratory Information Management Systems (LIMS), these systems might be upgraded to become sharable. Sharing LIMS data presents several challenges such as privacy and interoperability. Thus, sophisticated tools are needed to help users to submit meta-data and provide data in formats that facilitate data exchange, improve search, and reusability of models and data.

Addendum: Planning Document

Planning Document for the Quantitative and Systems Pharmacology (QSPcol) Workshop, Sept 25-26, 2008

Authors: QSPcol Consortium
Version: 20090223
Meeting website:http://meetings.nigms.nih.gov/meetings/QSPcolWorkshop/


This Planning Document for QSPcol was produced in advance of the Sept 25-26 meeting. The meeting abstract and announcement is given below. During the meeting, five breakout sessions were convened:

Breakout Session 1: Horizontal systems integration
Breakout Session 2: Vertical systems integration
Breakout Session 3: Quantitative biology and pharmacology
Breakout Session 4: Education and training
Breakout Session 5: Data management

Prior to the meeting, registered attendees were asked to provide written responses to the following areas

  • Generate a brief definition of the breakout session and contribute to a bibliography of key references, especially previous meeting reports that have significant import for the meeting.
  • Develop a list of issues, questions, and priorities for the breakout session.
  • Plan for developing future interactions between systems biology, pharmacology and drug discovery/action scientists.

The responses were edited and assembled into this document. The meeting website contains report-out from the breakout chairs and scribes.

Meeting Abstract

The Quantitative and Systems Pharmacology Workshop will provide state-of-the-art-knowledge and perspective about topics at the interface of systems biology and pharmacology to a highly diverse spectrum of researchers in academia, industry, and government. The question of where systems biology, modeling, and more quantitative measurements can be applied to pharmacology and drug discovery/action now and in the foreseeable future will be addressed. Accordingly, the talks will emphasize both conceptual information and significant research findings concerning the topic. The topic will be approached from the standpoint of both a horizontal integration (various networks in various cell systems) and a vertical integration (connections between pathways at different levels of integration, tissues, organs, etc.). Consideration of the state-of-the-art as well as future research needs in various areas will be made. The workshop will feature discussion sessions based on the presentations, throughout the two-day period. Meeting goals include highlighting where the state-of-the-art resides in relevant disciplines today, fostering integration of research efforts in these areas with others, identifying barriers (problems merging two disciplines), increasing collaboration between pharmacologists, clinicians, and systems biologists, and establishing pressing and long-term research needs (training, instrumentation, algorithms, etc) for advancement of the ability of systems biology to inform drug discovery and drug action.

NIH Program Staff: Michael Rogers (NIGMS), Sarah Dunsmore (NIGMS), Peter Lyster (NIGMS), Dick Okita (NIGMS), and Grace Peng (NIBIB)
Meeting Co-Chairs: William Jusko (U Buffalo), Douglas Lauffenburger (MIT)

Definitions and Plans for Breakout Sessions

Breakout Session 1: Horizontal Systems Integration

1. Proposed Definition

Systems integration is the act of assembling a composite system—computer models, in this case—from previously autonomous components used in specific contexts. Horizontal integration synthesizes a composite from components having the same spatial and/or temporal granularity.1 An example might be a cell pathway interaction model composed of various networks within and across different cells in various cell systems, but not including tissue or molecular dynamics. A clear statement of current and future uses to which an integrated system will be put is a precondition of systems integration for scientific research. A use statement typically begins with the current capabilities of the individual components followed by listing the expected capabilities of the integrated system.

2. Discussion Points from Comments Received

For horizontal systems development and integration, what are the expected uses for these types of models? How should such horizontal modeling approaches be integrated? Are there envisioned uses for a horizontally integrated composite model that are distinct from uses envisioned for a vertically integrated model?

What kinds of experimental measurements are required to provide parameters and input to horizontal models?

In mammalian systems, the cell and tissue behaviors that emerge during experiments are the consequences of local mechanisms—local component interactions. As integrated system models become more realistic and useful, should we anticipate being able to say the same of them? If so, how do biological components (at multiple levels) interact with each other. Alternatively, are there emergent properties of Multiscale phenomena (subcells, cells, tissue, organs, organism) which are not explained by the sum total of local mechanisms, and how can this emergent behavior be modeled.

It seems unlikely that biological component interactions will be exclusively either horizontal or vertical. Should we therefore anticipate that issues of horizontal and vertical component integration would merge into a single integration issue?

3. Characteristics

3.1 Data integration, data format, messaging, standards

How much of the modeling can be developed in a context-free manner, and what are the characteristics of the resulting implementation and software frameworks. Also, what are the approaches to handle the software needs of models that depend strongly on context.

Having a preliminary list of technologies that have succeeded or failed will help kick-start integration methodology. What data formats, formalisms, and standards have been successfully (or unsuccessfully) used for integration?

For software development, prioritize the usefulness of component interoperability versus component and system flexibility, adaptability, and reusability?

Because horizontally integrated systems consist of components having the same granularity, the data used to control, parameterize, and observe them will be semantically grounded at the same level. Doing so enables interoperability between components. Therefore, the format and protocol of data integration will rely on explicitly formulated ontologies and software and data exchange formats, e.g. XML or one of its derivatives like SBML. The same can be said of formalisms used to construct horizontally integrated systems. However, fixation on any given standard, formalism, or data format can result in component and/or system inflexibility, making it difficult to achieve new uses as the science advances.

Within the context of data integration, format, messaging, and standards, should we build and maintain competing and incommensurate standards, formats, and formalisms in order to promote and preserve technological agnosticism? What needs to be done to avoid fixating on any given standard, formalism, or data format? The risk of fixation may be greater for horizontally integrated systems than in vertically integrated systems because of the common semantic grounding due to a fixed spatial and temporal granularity.

Models of different organism components at the same granularity can have disparate data types and disparate models for how they interact with other components of the overall system. Is now the time to begin exploring methods for integrating these heterogeneous components into an overall system? An example of such a framework is the discrete event systems specification DEVS.

3.2 Exemplars, Uses, Capabilities

The ability to therapeutically target the molecular signaling and transcriptional pathways that drive cancer will be enhanced by a more global understanding of how these pathways interconnect to create, through feedback and cross-talk mechanisms, the full signaling network that integrates all signals into a net outcome or phenotype (e.g., oncogenic transformations). The broader biomedical research community is searching for the underlying rules that govern signaling, while cancer researchers are simultaneously addressing through technology development the need to measure variations between tumor types, between tumors from different patients, and even within tumors through single cell measurements (e.g., integrated microfluidic-based assays). Using cancer models such as Bcr-Abl driven leukemic transformation, we are collecting high dimensionality phosphoprofiling signaling data focused specifically on subnetworks that involve the cross-talk between a small number of signaling modules. Simplification of the problem through subnetwork analysis, allows us to first focus on a more tractable scale, while retaining clinical relevance, with the hope that we can later expand the network diversity using the technologies and methodologies we are developing. Through iterative rounds of global measurements, perturbation of the signaling proteins involved (e.g., drug inhibition of nodes and mRNA knock down techniques), and measurement of resultant phenotype, we are building experimentally grounded theoretical descriptions of the oncogenic systems, along with establishing the basis of pharmacological interventions.

What are the uses of horizontally integrated system models? Such systems will be used as stand-alone software components to study specific networks, and they will also be used as components in larger horizontally and vertically integrated systems. Such uses depend on the driving biological problems. If the integrated model is intended to represent an organism's pharmacological response, for example, then the duration of the response cycle, the number of cycles considered, and required response granularity become determining aspects. With that in mind, model and component reuse, flexibility, and adaptability, become important and that feeds back into model and component design. Therefore, it should be relatively easy to reconfigure components to represent different mechanistic hypotheses or different aspects of a key attribute under different experimental conditions. It should also be relatively simple to accommodate additional aspects at the current level of granularity or alter usage and assumptions, without requiring significant component or system reengineering. Components should be constructed so that they can be adapted easily to function as components in different, integrated models.

4. Issues, Questions, Needs, and Priorities

Are there envisioned uses for a horizontally integrated composite model that are distinct from uses envisioned for a vertically integrated model?

back to addendum menu

Breakout Session 2: Vertical Systems Integration

1. Proposed Definition

Vertical integration synthesizes a composite system using previously autonomous components that have different spatial and/or temporal granularities1. An example might be connecting cellular, tissues, organs, and whole body pathways. Clarity of uses and problems as well as application contexts are essential. Vertically integrated systems will need to be specified based on clear statements of usage, starting with the capabilities of the components, both in their original context and within the newly formed integrated system.

2. Discussion Points from Comments Received

For vertical systems development and integration, what are the expected uses for these types of models? How should such vertical modeling approaches be integrated? Are there envisioned uses for a vertically integrated composite model that are distinct from uses envisioned for a horizontally integrated model?

How do we obtain data sets that that drive scientific understanding of how molecular variations and defects lead to changed behavior at the tissue/organ level? Vertical systems integration will require non-invasive time series experiments that track molecular level dynamics and related phenotypic behavior, e.g., imaging-related studies that relate quantitative phosphoproteomics to proliferation rate and migration rate assays. Problems include different characteristic time scales (sec-min for proteins, hrs for migration, days for proliferation, weeks for in vivo, months-years for patients) and length scales (protein-cell-tissue), and thus the fact that, at least for the initial step, one will have to rely on pooling data from different experimental setups. Compounding the problem of assembling multiscale spatial images is that in vivo studies often suffer from resolution limits.

What approach do we take to couple models at different scales? Different models have different methods of synchronism in time, different database structures, and heterogeneous input and output data types. What kinds of enabling technologies should be addressed to address these issues e.g., the Ptolemy II modeling framework?

The promise of multiscale modeling is to assess the functional impact of molecular perturbations across the scales on interest. This will allow cross-scale biomarker assessment and thus may accelerate target validation for drug discovery research. This in turn should facilitate the entry of targets into the drug development pipeline and thus help focus treatment and reduce development costs. Clinical endpoints include decision support and treatment impact studies.

What capabilities should we expect of such systems and their components? Once integrated, it is unlikely that a system model will remain static. New information will require revisions, especially if that new information falsifies the model in some way. To answer new questions, the use statement may need revision. We anticipate needing to deconstruct the vertically integrated system, revise or replace components, and then assemble a revised system to undergo the next round of validation challenge. How can we make accomplishing these tasks easier? From an algorithm and software development perspective, one needs to avoid having to go through constant, and most definitely, costly revisions. A modular design may be needed, e.g., connecting nodes between pathway modules necessitates translation of code modules that can communicate, which in turn requires shared practices or standards. This is highly non-trivial for a number of reasons, including technical. It could be pushed as a community effort if one could agree on common standards—here is where the NIH could provide leadership.

Like any other computational modeling, these platforms should lead to experimentally testable hypotheses, should facilitate data integration and eventually enable outcome prediction (clinical translation). The latter is the really challenging part, and it relies in large parts on the data integration.

Handling tool sets at the front end should be as user-friendly as possible. This is a GUI design and not a platform issue, as the underlying algorithm will likely be quite sophisticated. Any of this requires constant support—again an area where the NIH could provide leadership. Hiring talent away from industry for non-profit money is difficult enough. Long-term sustainability is important especially where peer-review is used primarily for funding.

How will system models be falsified and/or validated? This will require experimental data on all levels, and models should be trained on clinical data. Construction and evaluation (selection, validation, falsification, and execution) are two fundamental aspects of building any system model. Each comes with many specific methods designed to maximize efficiency and efficacy. Can all necessary methods be assembled into a coherent methodology? Are the families of methods of construction distinct from those of evaluation?

How do we assess generalizability of models at each spatio-temporal scale?

How will we use these models directly for planning, monitoring, and adjusting therapy optimally? What types of stochastic control techniques are employed, and why?

Having increasingly implicit integration methods for heterogeneous components will make vertical integration of multi-attribute, biomimetic systems more straightforward. The components will be models in their own right. Vertical integration will be made easier when multiple, heterogeneous models are able to operate simultaneously within a common simulation framework. Coupling within such a framework will require an automated means of system behavior evaluation during execution. To achieve this, data standards for data exchange as well as data characterization and classification are needed. Additionally, maintaining data quality through filtering techniques is critical.

Other issues: scalability; development of model sharing environments; develop of widely-used ontologies, IP issues; grid and high-performance computing access; and workflow design, storage and execution.

3. Characteristics

3.1 Data integration, data format, messaging, standards

How much of the modeling can be developed in a context-free manner, and what are the characteristics of the resulting implementation and software frameworks. Also, what are the approaches to handle the software needs of models that depend strongly on context.

Can ontologies be used for integrating models at different levels (e.g. Gennari et al, 2008)? Are existing formats and standards (e.g. SBML, CellML) adequate for multiscale modeling? Having a preliminary list of technologies that have succeeded or failed will help kick-start integration methodology. What data formats, formalisms, and standards have been successfully (or unsuccessfully) used for integration?

We should expect that some details of vertical systems integration will be unique to the specific usage context, and that usage contexts will evolve and change. Consequently, we need to insure flexibility, adaptability, and reusability. For example, because it is infeasible to include an equivalent amount of detail in a sub-cellular model, as in a reasonably realistic organ model, we can anticipate that a vertically integrated system composed of both sub-cellular and organ sub-systems may require different patterns of messaging, data formats, formalisms, and standards for construction and evaluation. Does it follow that we need to avoid restricting to any given technology, including data formats, standards, or formalisms?

How do we characterizing usage and problem contexts for vertically integrated system and their components as a whole?

3.2 Exemplars, Uses, Capabilities

The In Silico Liver (ISL) exhibits elements of both vertical and horizontal integration. It uses an In Silico Hepatocyte (ISH, Yan et al. 2008) model as one of its sub-systems. The two separate aspects (use cases) are: 1) in situ perfusion output fraction profiles for compounds studied the ISL and 2) uptake by cultured hepatocytes of compounds for the ISH.

Multi-formalism modeling tools can facilitate the synthesis of vertically integrated systems. For example, the Ptolemy II software framework allows one to assemble graphs where the nodes are software objects (or scripts) containing anything expressible in software (including network- or database-enabled callouts) and the edges are interactions between them. The semantics of the graph is added through various different "Directors," which specify what type of data goes across the edges and how the nodes interact (e.g. data-driven versus event-driven). Ptolemy's facility for specifying composite nodes, themselves governed by different "Directors," demonstrates its usefulness for the synthesis of hierarchical, multi-scale, and multi-formalism models. An example of such a model is provided by McPhillips et al. 2006, wherein different databases are assembled into a systemic workflow.

4. Issues, Questions, Needs, and Priorities

  1. Quantitative data sets for (a) concentrations and sites of cellular components and (b) kinetic rate constants that enable the building of quantitative models onto which the pharmacokinetic (PK) and pharmacodynamic (PD) data can be mapped.
  2. Reasonably comprehensive physiological data sets in two or three model areas. This first needs a gap analysis to estimate how these data sets can be obtained from the published data and how much needs to be gathered denovo
  3. Genome variation—environment interactions
  4. Comparing simulations of "perturbed" vs. "reference" systems. There is a need tools for comparing simulations for perturbations such as genetic variation or comparing different qualitative or quantitative interventions
  5. Study the impact of microenvironmental factors on cell and multi-cell systems, with quantitative phosphoproteomics that are connected to phenotypic assessments
  1. Trustability—Should the approach taken to construct, evaluate, and build trust in vertically integrated systems be analogous to the approach taken in developing, evaluating, and using wet-lab experimental systems? Given the nature of software, there is strong, technical and scientific justification for simulation scientists to understand, from top to bottom, the implementation and execution of models with which they work. Achieving that understanding usually requires being able to implement the model from scratch. The same is true for wet-lab experimentation. However, the scientific community discourages "reinventing the wheel" for wet-lab experimentation by focusing on methods and repeatability, and on cataloging and documenting the components of experiments (including labware, living parts, and key materials along with instruments used for measurement, analysis, visualization, etc.), rather than expecting all scientists to understand (and be able to create) all components from the bottom up. To increase the pace of scientific progress with simulation software, might we benefit from developing and establishing similar methods for establishing trust in components and in vertically integrated systems? We can anticipate the above problem becoming particularly important in vertically integrated systems synthesis because we will expect components to be acquired from different sources. The contents of each component cannot be painstakingly analyzed by each user or for each context.Within other domains, computer science for example, what methods have been successful in establishing trust in heterogeneous components having diverse origins? Does the use of open-source (or open-standard) tools facilitate or debilitate the establishment of trust?
  2. Dynamic systems of systems execution—Should we strive for a methodology to facilitate dynamic execution in vertically integrated systems? Experience in other domains has taught that any vertically integrated system, particularly when it is multi-formalism, will present a complexity bottleneck at some point. Even if the system can be parallelized (in process and memory), there is still a heuristic value bottleneck, because the system can become so complex as to require specialists to manage and understand it. Modularization is one approach.
  3. Automated construction and evaluation—Can we identify which methods and tools have facilitated the automatic construction and evaluation of vertically integrated systems within other domains? Having that information might facilitate identification of key capabilities and technologies needed to do the same within the biomedical domain.

back to addendum menu

Breakout Session 3: Quantitative Biology and Pharmacology

1. Proposed Definition

Systems biology provides a set of tools for specific tasks in drug discovery and development. The applications discussed in this break out group may begin at a stage when a target has been selected. This selection rationale is based on a varying foundation of mechanistic understanding. Examples of such rationales are:

  1. a knock-out animal model showing a desirable phenotype
  2. an experimental observation of a genetic mutation in a disease-resistant human population
  3. a mechanistic hypothesis of disease etiology
  4. selection from a whole genome screen combined with literature support

Technical, legal and business considerations also factor into progressing a target. At the discovery stages of the drug development pipeline, there is considerable research on the chemistry of modifying the behavior of the target using a small molecule. Hidden in the pipeline is the work of optimizing the chemistry to make a compound that is stable, bioavailable, safe, manufactured on a large scale, and formulated for delivery. From a modeling standpoint this is a perfect example of a multiscale modeling problem. From a quantitative biology standpoint, the following table suggests a number of places where systems biology impacts the needs of the process.

Select back-up targetsIdentification of alternative ways to modulate a target of interest; development of the topology and connectivity of the system; study of intermediate connections
Predict strategy for therapeutically modulating target1. Determine functional relationships from qualitative effects (e.g. of inhibition and excitation)
2. Predict variations in response due to variations in the population
3. Predict genetic and biochemical biomarkers for efficacy
Select assay(s) for screening, lead optimization, and validationCue-signal-response study to determine which endpoints are most informative
Predict on-target safetyIdentify potential "target-related off-target" effects by identifying how the target contributes to known toxicity pathways and mechanisms of diseases not under study
Therapeutic re-purposingPredict potential alternative diseases and/or drug combinations based on known mechanisms

Quantitative and systems pharmacology is the interface or boundary between drug discovery and development. Modeling and systems approaches are already being done on the development side (e.g., see 2008 meetings ACoP in Tucson and AAPS NCB in Toronto). An ideal work flow in quantitative systems pharmacology might be:

  1. Development of the topology and connectivity of the system.
  2. Study of the intermediate connections.
  3. Determine functional relationships from qualitative effects (e.g. of inhibition and excitation).
  4. Quantification of the functional relationships.
  5. Study modulating effects occurring through disease, natural aging, and pharmaceuticals.
  6. Determine appropriate molecular targets and molecules to move forward to drug development.
  7. Simulate clinical study designs based on integrated understanding of system dependencies.
  8. Determine how best to use the models for therapy at the bedside.

2. Discussion Points from Comments Received

  1. Systems approaches might be used to identify "off target" mechanisms of drug action.
  2. Systems approaches might be projected from the study of drug action in cellular models in vitro to replication ex vivo in model systems and ultimately in humans.
  3. How we might begin to integrate systems analysis of diverse outputs of drug response in a single model system.
  4. How pharmacologic and genetic manipulation of a single pathway might be utilized to interrogate a more general response (e.g., PG pathway manipulation to interrogate the inflammatory response).
  5. Inter (or indeed intra) individual variation in systems perturbation by the same pharmacological intervention.
  6. In addition to the idealized pipeline (discovery to development to commercialization), we should consider the important role played by "bedside to bench" development. An historical example of this would be the development of allopurinol for gout by Wayne Rundles at Duke. Technically, this might be considered an "off target" indication.
  7. One needs to consider the development and qualification of biomarkers as well as drugs. In many cases, biomarkers will more easily lend themselves to incorporation in quantitative models than downstream drug effects. Their investigation will also provide important leads in pathway development.
  8. It seems likely that different approaches will be appropriate for different drug classes. Considerable relevant information can be provided by in vitro studies and models of antibiotic action than is the case for antipsychotics.
  9. Studies that incorporate the modulating effects of pharmacologic agents are conducted with greater frequency than is implied in the document. Obviously, the granularity of these studies has usually not been very fine.
  10. There needs to be greater emphasis on clinical pharmacology. If you look at a drug development scheme, the areas that are highlighted are those in which clinical pharmacologists have particular expertise. There is an obvious gap in clinical pharmacology education, and the NIH course and textbook are a good start.
  11. Relevance of normal versus induced and/or natural disease conditions.
  12. Some drug discovery is based on empirical, trial and error studies that are in turn based on historical trial and error studies.

3. Exemplars

  1. Insulin resistance diabetes, obesity, and atherosclerosis, e.g., there are drugs such as thiazolidinediones (TZD) used for insulin-based approaches.
  2. Hypertension; there are good drugs affecting hypertensive pathways and good therapeutics, e.g., diuretics.
  3. Cancer, e.g., breast and ovarian cancer, e.g. tamoxifen.
  4. Inflammation, e.g., there are anti-inflammatory drugs, e.g. cox2 inhibitors.
  5. Bone...
  6. Digestive disorders...
  7. Infectious diseases...

back to addendum menu

Breakout Session 4: Education and Training

1. Definition and Challenges

Defining the new combinations of competencies, masteries, habits of mind and interactive skills required by the future investigators who will successfully use systems approaches to understand diseases and address problems of drug development. Defining the pedagogies, curricula and activities that will constitute training for these new combinations.

Pedagogical challenges

Teaching (medical, pharmacy, as well as graduate students) thinking about biological systems (where vertical and horizontal integration are not constant) as distinct from engineering systems theories (where organization is fixed; and where the theory does not map well to biology) is a major challenge. The following are several reasons:

  1. The novelty of the field; concepts and ideas are still emerging, and those ideas are necessarily very different from the hyper-reductionist thinking that still dominates biomedical training.
  2. The field has a inter-, intra-, and multidisciplinary nature.
  3. The field has a diverse research community and literature
  4. The field has a lack of:
    • a common language, consequently the need (for individuals in an unorganized fashion) to borrow language from other fields (research in different domains have different meanings for the same terms).
    • well-defined subject boundaries.
    • generally accepted definitions; the absence of methods that are designed to achieve a long term vision.
    • understanding and willingness in industry (and the FDA and NIH) to support such research.
  5. Collaboration is needed between government, industry, and academia (medical and pharmacy schools) to further develop this field. With the Critical Path initiative, the FDA is trying to promote more dynamic science-driven pre-clinical and clinical drug development paradigms. Their efforts in pharmacometrics have been significant as well.
  6. Instruction in special courses and modules requires multi-domain backgrounds.
  7. Potential newcomers to the field (those initially attracted to the issues and problems) find it difficult to change their thinking.
  8. The familiarity and consistency of a "domain-centered culture" is absent (it is multidisciplinary); there is no single theory (it is multiparadigm [multi-attribute; multi-perspective]).
  9. Both depth and breadth are pursued concurrently
  10. There is a need to use techniques of stochastic adaptive control to optimize therapy for patients, not just for research.

2. Discussion Points from Comments Received

Do we need integrated courses both Pharmacy and Systems Biology grad students and MDs who would like to have a career in Systems Pharmacology and Therapeutics?

What further training in quantitative pharmacology is needed by those in Systems Biology to appreciate the intricacies of drug disposition and action?

Discuss the needs and barriers. Design educating and training programs for students in Quantitative and Systems Pharmacology:

  1. What type of training is needed (MS, PhD, Postdoctoral Fellowship)?
  2. What type of student should be recruited?
  3. What coursework should be required (quantitative sciences; laboratory sciences)?
  4. What hands-on modeling experience is needed?
  5. How will programs be funded?
  6. What types of programs should be involved?
  7. How will programs develop a core of qualified faculty?
  8. Should a consortium approach be considered since data may reside with industry partners?
  9. How to pursue both depth and breadth concurrently and keep the balance?
  10. How to prepare trainees for both academic and industrial jobs?

Demand vs. supply analysis:

  1. How much will the demands be from both academia and industry in the near term and long term?
  2. How many students are we expecting to graduate in the near future with the right skill set?
  3. Do we need to expand our current education programs and by how much?
  4. If so, what are the challenges to expand the program (e.g. funding, teachers, attract potential students etc.)?

Defining the new combinations of competencies, masteries, habits of mind and interactive skills required by the future investigators who will successfully use systems approaches to address problems of drug development and drug interactions. Defining the pedagogies, curricula and activities that will constitute training for these new combinations.

One of the major limitations for a more widespread adoption in the scientific community, and in particular in the pharmaceutical industry and clinicians, is a great shortage of appropriately trained clinicians, pharmacists, and junior scientists that have sufficient expertise in this area and the associated techniques. The focus of the Group 4 breakout session should be to collect ideas and strategies on how to overcome this shortage. Over the last two decades, there has been a steady decline in academic programs and training sites that have traditionally provided or could provide education and training in quantitative pharmacology, predominantly programs in pharmaceutical sciences and biomedical engineering. This decline may be due to the reluctance and/or incapability of clinicians to adjust to these new approaches, and to understand them, to use them, and to teach them. The greatest challenge today may lie in creating new training sites by attracting and recruiting junior faculty and convincing academic administrations that investing in these junior positions is promising and important. Attracting external funding seems to be the greatest concern. Quantitative pharmacology as a translational science has for a long time not been a focus of federal funding, and many academic administrators as well as junior faculty still perceive it that way. It will be crucial for the success and development of quantitative pharmacology as a discipline to reverse this trend and establish a large and diverse number of educational programs and training sites that instruct a new generation of translational scientists with focus on quantitative and systems pharmacology.

Discuss how we can better prepare the next generation of biologists and clinicians for the increasing demands of quantitative and systems pharmacology. The session may be used to discuss what types of skills are needed as a quantitative and systems pharmacologist; how schools are training their students with those skills; what are the challenges during the training; how we can improve the training program. In addition, how much will the demands be from both academia and industry in the near term and long term; how many students are expected to graduate in the near future with the right skill set; is it necessary to expand our current education programs; if so, what are the challenges to expand the program (e.g. funding, teachers, attract potential students etc.)?

Funding issues:

  1. Government funding
  2. Industrial funding

3. Courses, training grants, environment

In this session, there should be a discussion on the need to expand current education program and if so, how big the expansion needs to be; to identify the biggest challenges for such training program; and to propose potential solutions for the challenges.

Environment: What may be required is a major research university with a comprehensive health affairs campus that includes a teaching hospital, translational research/clinical trials unit, availability of requisite support sciences (bioinformatics, biostatistics, genetics/genomics) and a strong relationship with the pharmaceutical industry and relevant federal agencies that can provide appropriately interactive training experiences for both graduate students and fellows would be required.

4. Goals, Issues, questions, needs, and priorities


  • To provide academic training in:
    • PK/PD modeling and simulation
    • The application of modeling and simulation to clinical trial design and the development of individualized PK/PD/PG dosing algorithms
    • Disease process/disease progression modeling
    • The development of PK/PD models informed by mechanistic experimentation
    • The integration of biomarker strategies into PK/PD profiling


  • Human Pharmacology Core Coursework
    • Pharmacokinetics/Pharmacodynamics
      • Biopharmaceutics
      • Drug Metabolism and Transport
      • Pharmacogenomics
    • Experimental/Clinical Pharmacology and Toxicology
  • Pharmacometrics Core Coursework
    • Population Pharmacokinetics
    • Clinical Trial Simulation and Experimental Design
    • Bayesian Methods
  • Statistics Core Coursework
    • Applied Biostatistics
    • Advanced Statistical Methods
      • Regression Analysis
      • Multivariate Analysis
      • Longitudinal Data Analysis
      • Survival Analysis
    • Monte Carlo Methods
  • Programming Core Coursework
    • Computational Methods/Application and Development
    • Introduction to Statistical Programming
  • Hands-on Modeling and Simulation Projects

back to addendum menu

Breakout Session 5: Data Management

1. Definition and Challenges:

Data management involves three concepts: data formats for exchange; data messaging for communication; data storage. Special attributes of biomedical data are: heterogeneous data types; data acquisition pipeline; imperfect laboratory information management systems (LIMS); open and closed databases and tools.

2. Discussion Points from Comments Received

Systems Pharmacology has particular challenges for data analysis and management because it involves the union of previously disparate informatics disciplines. First, the simulation and modeling aspects of systems biology are clearly the most relevant and applicable aspects of this work. But they require an informatics infrastructure that may be more diverse than other fields. Chemical informatics is critical in describing the structures and activities of small molecules. Bioinformatics focuses on the genes and protein products that are measured in the genome, the transcriptome and the proteome. Physiological modeling is important to connect the molecular scale to organs as an enabling tool for PK/PD. Finally, clinical informatics studies how clinical data can be organized and mined for significant phenotypic trends. At the same time, there are special, new data sources that may be specific to systems pharmacology. The goal of this breakout group is to answer the following questions:

  1. What are the new data sources from systems pharmacology that need to be collected and analyzed?
  2. What are the basic algorithms required for organizing, storing, retrieving and analyzing systems pharmacology data?
  3. What key capabilities of chemical informatics are required for systems pharmacology? Are they available to researchers today?
  4. What key capabilities of biomedical informatics and computational biology are required for systems pharmacology? Are they available to researchers today?
  5. What key capabilities of clinical informatics are required for systems pharmacology? Are they available to researchers today?

Potential and challenges: Integration and scalability. Scalability (to clinically relevant levels), works towards multi-scale, multi-resolution modeling & simulation. Integration puts a focus on interoperability, ontologies, data storage, transmission, and formats. A question is: How are clinical trials that intend to address personalized medicine set up in the future if population-based assessments are deemed obsolete? Computational biology promises progress in the early stages of the pipeline and on diagnostics side first but it is also important to think early about how this possibly affects treatment, testing and clinical routine later on.

3. Goals, Issues, Questions, Needs, and Priorities

To discuss statistical and bioinformatic models for horizontal and vertical data integration, and to discuss approaches towards public access databases and parallelization of data interpretation.

Genome-enabled data sets as applied to pharmacogenomics systems biology include large data sets on multiple time scales. After treatment with a drug, multiple changes occur at the molecular level, including protein phosphorylation within seconds, proteomic profiling changes in localization occur in minutes, alteration of stored RNAs for translation within minutes, and transcriptional changes within hours. Serum markers and physiological changes can mirror any of these lower level molecular alterations, with added complications of organ-organ effects. It is now possible to 'profile' both low level molecular responses to a drug, and higher level reactions to these responses, resulting in a multi-scale physiome model of pharmacology.

Ideally, all levels and types of data from a specific drug response can be integrated and analyzed as unit, and cause/effect models able to predict drug responses evolved. Also, these large integrated data sets should be made public, so that the highly complex and innovative data analyses required for model development can be parallelized between many scientists worldwide.

This presents challenges for both data integration, databasing, and public access. Also, the statistical and bioinformatic tools required to evolve multi-scale physiome models of pharmacology are in their infancy. One approach that has emerged in the informatics community is developing workflows where tools and datasets are chained such that the output from one tool feeds into the input of another tool.

Data storage capacity and high-performance computation are two important components that should be addressed as well in this discussion. Current trend suggests that it is more economically feasible to centralize data storage as well as data processing. Google's success is the motivation behind this viewpoint.

More efforts should be put into data organization. The design of a global Systems Pharmacology data warehouse that would mine, organize, disseminate data in a cloud computing fashion is one potential major useful undertaking.

back to addendum menu


Meetings and Meeting Summaries

AAPS National Biotechnology Conference. 2008 Jun 22-25; Toronto, Ontario, Canada. Meeting website: http://www.aaps.org/meetings/biotec/bt08/index.asp .

Aegean conference on pathways, networks, and systems. Greece. Meeting website: http://www.aegeanconferences.org/

The American Conference on Pharmacometrics. 2008 Mar 9-12; Tucson, Arizona. The 2A and 2B sessions on the major ones on Systems Biology.

The American Society for Pharmacology and Experimental Therapeutics. Meeting website: http://www.aspet.org/public/meetings/meetings.html

Data Integration in the Life Sciences. Meetings website

Krishna R, Schaefer HG, Bjerrum OJ editors. Effective integration of systems biology, biomarkers, biosimulation and modelling in streamlining drug development. Meeting report: EUFEPS Conference on Optimising Drug Development together with European Biosimulation Network of Excellence: Biosimulation - A New Tool in Drug Development. Eur J Pharm Sci. 2007 May;31(1):62-7. Epub 2007. Cited in: PubMed; PMID: 17408933. Meeting website: http://www.eufeps.org/document/con_basel_nov06.html

Fifth Symposium Functional Genomics Critical Injury and Illness. 2007 Nov 14-15; Natcher Auditorium, Bethesda, MD. Meeting website: http://www.strategicresults.com/fg5/

Forum on Modeling and Simulation Applications in Clinical Pharmacology (MoSAiC). 2008 Mar 9-12; Bridgewater, New Jersey. Meeting website: http://www.mosaicnj.org/

International Workshop on Uncertainty and Variability in Physiologically Based Pharmacokinetic (PBPK) Models. 2006 Oct 31-Nov 2; Research Triangle Park, North Carolina. ?

Rowland M, Balant L, Peck C. Physiologically Based Pharmacokinetics in Drug Development and Regulatory Science: A Workshop Report. 2002 May 29-30; Georgetown University, Washington, DC. AAPS PharmSci. 2004;6(1):1-12. Article 6; Cited in: AAPSJ DOI: 10.1208/ps060106. Available from: http://www.aapsj.org/view.asp?art=ps060106

RECOMB International Conference on Research in Computational Molecular Biology. Meetings website: http://www.recomb2007.com/html/previousConf.html

D'Argenio DZ, organizer. Workshop on Advanced Methods of PK/PD Systems Analysis. 2007 Jun 15-16; Biomedical Simulations Resource. Available from: http://bmsrs.usc.edu/Service.Training/workshops/files/2007.pdf

Review Articles and Books

Allen JE, Gardner SN, Slezak TR. DNA signatures for detecting genetic engineering in bacteria. Genome Biology. 2008;9(3) 9:R56. Cited in: PubMed; PMCID: PMC2397508.

Atkinson AJ Jr, Abernethy DR, Daniels CE, Dedrick R, Markey SP. Principles of Clinical Pharmacology. Second ed. Elsevier; 2007. 545 p. Principles of Clinical Pharmacology; NIH Clinical Center Course. Meeting website: http://www.cc.nih.gov/training/training/principles.html

Chabot J, Gomes B. Modeling efficacy and safety of engineered biologics. 2009; Ekins S, Xu JJ, editors. Drug efficacy, safety, and biologics discovery: emerging technologies and tools. Wiley and Sons series on technologies for the pharmaceutical industry.

Hendriks BS. Applications of systems biology approaches to target identification and validation in drug discovery. 2009; Ekins S, editor. Systems biology in drug discovery and development. Wiley and Sons.

Landersdorfer CB, Jusko WJ. Pharmacokinetic/Pharmacodynamic Modelling in Diabetes Mellitus. Clinical Pharmacokinetics. 2008; 47(7):417-448. Cited in: PubMed; PMID: 18563953.

Recommendations of the National Commission on Digestive Diseases. Opportunities and Challenges in Digestive Disease Research: Recommendations of the National Commission on Digestive Diseases. 2007-2008; Available from: https://www.niddk.nih.gov. National Commission on Digestive Diseases site: https://www.niddk.nih.gov

US EPA National Center for Computational Toxicology (NCCT). A Framework for a Computational Toxicological Research Program. 2003; Available from: http://nepis.epa.gov/Exe/ZyPURL.cgi?Dockey=100046MA.t​xt

Journal Articles

Albeck JG, MacBeath G, White FM, Sorger PK, Lauffenburger DA, Gaudet S. Collecting and organizing systematic sets of protein data. Nat Rev Mol Cell Biol. 2006 Nov;7(11):803-12. Cited in: PubMed; PMID: 17057751.

An G. Concepts for developing a collaborative in silico model of the acute inflammatory response using agent-based modeling. J Crit Care, 2006; 21:105-111. Cited in: PubMed; PMID: 16616634.

Andersen ME. Toxicokinetic modeling and its applications in chemical risk assessment. Toxicol Lett. 2003 Feb 18;138(1-2):9-27. Cited in: PubMed; PMID: 12559690.

Atkinson AJ, Lalonde RL. Introduction to quantitative methods in pharmacology and clinical pharmacology: A historical overview. Clin Pharmacol Ther. 2007;82(1):3-6. Cited in: PubMed; PMID: 17571065.

Barrett JS, Fossler MJ, Cadieu KD, Gastonguay MR. Pharmacometrics: a multidisciplinary field to facilitate critical thinking in drug development and translational research settings. J Clin Pharmacol. 2008;48:632-649. Cited in: PubMed; PMID: 18440922.

Barton HA, Chiu WA, Woodrow Setzer R, Andersen ME, Bailer AJ, Bois FY, Dewoskin RS, Hays S, Johanson G, Jones N, Loizou G, Macphail RC, Portier CJ, Spendiff M, Tan YM. Characterizing uncertainty and variability in physiologically based pharmacokinetic models: state of the science and needs for research and implementation. Toxicol Sci. 2007 Oct;99(2):395-402. Cited in: PubMed; PMID: 17483121.

Bayard D, Jelliffe R: A Bayesian Approach to Tracking Patients having Changing Pharmacokinetic Parameters. J. Pharmacokin. Pharmacodyn 2004;31(1):75-107. Cited in: PubMed; PMID: 15346853.

Bleyzac N, Souillet G, Magron P, Janoly A, Martin P, Bertrand Y, Galambrun C, Dai Q, Maire P, Jelliffe R, Aulagner G: Improved clinical outcome of paediatric marrow recipients using a test dose and Bayesian pharmacokinetic individualization of busulfan dosage regimens. Bone Marrow Transplantation. 2001;28:743-751. Cited in: PubMed; PMID: 117816025.

Bustad A, Terziivanov D, Leary R, Port R, Schumitzky A, Jelliffe R. Parametric and nonparametric population methods: their comparative performance in analyzing a clinical dataset and two Monte Carlo simulation studies. Clin Pharmacokinet. 2006;45(4):365-383. Cited in: PubMed; PMID: 16584284.

Cho CR, Labow M, Reinhardt M, Van Oostrum J, Peitsch MC. The application of systems biology to drug discovery. Curr Opin Chem Biol. 2006;10:294-302. Cited in: PubMed; PMID: 16822703.

Corvasier S, Maire P, Bouvier d'Yvoire M, Barbaut X, Bleyzac N, Jelliffe R. Comparisons between Antimicrobial Pharmacodynamic Indices and Bacterial Killing as Described by Using the Zhi Model. Antimicrobial Agents and Chemotherapy. 1998;42:1731-1737. Cited in: PubMed; PMCID: PMC105675.

Danhof M. Mechanism-Based PD Modeling for Predicting Exposure Response. The American Conference on Pharmacometrics. 2008 Mar 9-12; Tucson, Arizona.

Deisboeck T. References: multi-scale cancer modeling.

Deisboeck TS, Zhang L, Yoon J, Costa J. In silico cancer modeling: is it ready for primetime? Nature Clinical Practice Oncology. 2009 Jan;6(1):34-42. Cited in: PubMed; PMID: 18852721.

Di Bernardo D, Thompson MJ, Gardner TS, Chobot SE, Eastwood EL, Wojtovich AP, Elliott SJ, Schaus SE, Collins JJ. Chemogenomic profiling on a genome-wide scale using reverse-engineered gene networks. Nat Biotechnol. 2005 Mar;23(3):377-83. Cited in: PubMed; PMID: 15765094.

Dolinski K, Botstein D. Changing perspectives in yeast research nearly a decade after the genome sequence. Genome Res. 2005;15:1611-1619. Cited in: PubMed; PMID: 16339358.

Earp JC, Dubois DC, Molano DS, Pyszczynski NA, Keller CE, Almon RR, Jusko WJ. Modeling corticosteroid effects in a rat model of rheumatoid arthritis I: mechanistic disease progression model for the time course of collagen-induced arthritis in Lewis rats. J Pharmacol Exp Ther. 2008 Aug;326(2):532-45. Cited in: Pubmed; PMCID: PMC2574807.

Earp JC, Dubois DC, Molano DS, Pyszczynski NA, Almon RR, Jusko WJ. Modeling corticosteroid effects in a rat model of rheumatoid arthritis II: mechanistic pharmacodynamic model for dexamethasone effects in Lewis rats with collagen-induced arthritis. J Pharmacol Exp Ther. 2008 Aug;326(2):546-54. Cited in: Pubmed; PMCID: PMC2574741.

Edginton AN, Theil FP, Schmitt W, Willmann S. Whole body physiologically-based pharmacokinetic models: their use in clinical drug development. Expert Opin Drug Metab Toxicol. 2008 Sep;4(9):1143-52. Cited in: PubMed; PMID: 18721109.

Fisher J, Henzinger TA. Executable cell biology. Nat Biotech. 2007;25(11):1239-49. Cited in: PubMed; PMID: 17989686.

Forst CV. Host-pathogen systems biology. Drug Discovery Today. 2006;11(5/6):220-7. Cited in: PubMed; PMID: 16580599.

French ED. Academic pharmacologists: Confronting new challenges in educational programs of graduate and health care professionals. J Pharmacol Exper Ther. 2004;309:441-443. Cited in: PubMed; PMID: 14766945.

Gardner TS, di Bernardo D, Lorenz D, Collins JJ. Inferring genetic networks and identifying compound mode of action via expression profiling. Science. 2003 Jul 4;301(5629):102-5. Cited in: PubMed; PMID: 12843395.

Garmire LX, Garmire DG, Hunt CA. An in silico transwell device for the study of drug transport and drug-drug interactions. Pharm Res. 2007;24(12):2171-86. Cited in: PubMed; PMID: 17703347.

Garmire LX, Hunt CA. In silico methods for unraveling the mechanistic complexities of intestinal absorption: metabolism-efflux transport interactions. Drug Metab Dispos. 2008;36(7):1414-24. Cited in: PubMed; PMID:16353926.

Gennari JH, Neal ML, Carlson BE, Cook DL. Integration of Multiscale biosimulation models via light-weighted Semantics. Pacific Symposium on Biocomput. 2008;13:414-425. Available from: http://psb.stanford.edu/psb-online/proceedings/psb08/gennari.pdf

Grasela TH, Fiedler-Kelly J, Walawander CA, Owen JS, Cirincione BB, Reitz KE, Ludwig EA, Passarell JA, Dement CW. Challenges in the transition to model-based development. Aaps J. 2005;7:E488-495. Cited in: PubMed; PMID: 16353926.

Hellerstein MK. Exploiting complexity and the robustness of network architecture for drug discovery. J Pharmacol Exper Ther. 2008;325:1-9. Cited in: PubMed; PMID: 18202293.

Holford N, Karlsson MO. Time for quantitative clinical pharmacology: A proposal for a pharmacometrics curriculum. Clin Pharmacol Ther. 2007;82:103-105. Cited in: PubMed; PMID: 17495873.

Hunt CA, Ropella GEP, Yan L, Hung DY,Roberts MS. Physiologically-based synthetic models of hepatic disposition. J Pharmacokinet Pharmacodyn. 2006;33:737-72. Cited in: PubMed; PMID: 17051440.

Hunt CA, Ropella GEP, Park S, Engelberg J. Dichotomies Between Computational and Mathematical Models. Nat Biotech. 2008;26:757-9. Cited in: PubMed; PMID: 18612289.

Ideker T, Winslow LR, Lauffenburger DA. Bioengineering and systems biology. It endeavors to articulate a bioengineering perspective on systems biology and includes some information on relevant educational programs. Ann Biomed Eng. 2006 Jul;34(7):1226-33. Cited in: PubMed; PMID: 16929563.

Janes KA, Kelly JR, Gaudet S, Albeck JG, Sorger PK, Lauffenburger DA. Cue signal-response analysis of TNF-induced apoptosis by partial least squares regression of dynamic multivariate data. J Comput Biol. 2004;11(4):544-61. Cited in: PubMed; PMID: 15579231.

Janes KA, Lauffenburger DA. A biological approach to computational models of proteomic networks. Curr Opin Chem Biol. 2006;10(1):73-80. Cited in: PubMed; PMID: 16406679.

Jelliffe R. Estimation of Creatinine Clearance in Patients with Unstable Renal Function, without a Urine Specimen. Am. J. Nephrology. 2002;22:320-324. Cited in: PubMed: 12169862.

Jelliffe R, Schumitzky A, Bayard D, Van Guilder M, Leary R, Botnen A, Gandhi A, Maire P, Barbaut X, Bleyzac N, Bondareva I, Neely M. Pharmacokinetic Methods for Analysis, Interpretation, and Management of tdm Data, and for Individualizing Drug Dosage Regimens Optimally. Hempel G, editor. Handbook of Analytical Separations. 2004;5:Drug Monitoring and Clinical Chemistry:129-168.

Krishnan K, Johanson G. Physiologically-based pharmacokinetic and toxicokinetic models in cancer risk assessment. J Environ Sci Health C Environ Carcinog Ecotoxicol Rev. 2005;23(1):31-53. Cited in: PubMed; PMID: 16291521.

Kumar N, Hendriks BS, Janes KA, De Graaf D,Lauffenburger DA. Applying computational modeling to drug discovery and development. Drug Discov Today. 2006 sep;11(17-18):806-11. Cited in: PubMed; PMID 16935748.

Lalonde RL, Kowalski KG, Hutmacher MM, Ewy W, Nichols DJ, Milligan PA, Corrigan BW, Lockwood PA, Marshall SA, Benincosa LJ, Tensfeldt TG, Parivar K, Amantea M, Glue P, Koide H, Miller R. Model-based Drug Development. Clin Pharmacol Ther. 2007;82:21-32. Cited in: PubMed; PMID: 17522597.

Lavé T, Parrott N, Grimm HP, Fleury A, Reddy M. Challenges and opportunities with modelling and simulation in drug discovery and drug development. Xenobiotica. 2007 Oct-Nov;37(10-11):1295-310. Cited in: PubMed; PMID: 17968746.

Levin JM, Penland RC, Stamps AT, Cho CR. Using in silico biology to facilitate drug development. Novartis Found Symp. 2002;247:222-38. Cited in: PubMed; PMID: 12539958.

Macdonald I, Staatz C, Jelliffe R, Thomson A. Evaluation and Comparison of Simple Multiple Model, Richer Data Multiple Model, and Sequential Interacting Multiple Model (IMM) Bayesian Analyses of Gentamicin and Vancomycin Data Collected From Patients Undergoing Cardiothoracic Surgery. Ther. Drug Monit. 2008;30:67-74. Cited in: PubMed; PMID: 18223465.

Mager DE, Jusko WJ. Development of translational pharmacokinetic-pharmacodynamic models. Clin Pharmacol Ther. 2008;83:909-912. Cited in: PubMed; PMID: 18388873.

Mager DE, Wyska E, Jusko WJ. Diversity of mechanism-based pharmacodynamic models. Drug Met Dispos. 2003;31:510-518. Cited in: PubMed; PMID: 12695336.

Mani K, Lefebvre C, Wang K, Lim WK, Basso K, Dalla-Favera R, Califano A. A Systems Biology Approach to the Prediction of Causal Oncogenic Mechanisms and Drug Mechanism-of-Action Profiles in Cancer Phenotypes. Molecular Systems Biology. 2008;4:169. Cited in: PubMed; PMID: 18277385.

Martin P, Bleyzac N, Souillet G, Galambrun C, Bertrand Y, Maire P, Jelliffe R, Aulagner G. Relationship between CsA trough blood concentration and severity of acute graft-versus-host disease after paediatric stem cell transplantation from matched sibling or unrelated donors. Bone Marrow Transplantation. 2003;32:777-784. Cited in: PubMed; PMID: 14520421.

Materi W, Wishart DS. Computational systems biology in drug discovery and development: methods and applications. Drug Discovery Today. 2007;12:295-303. Cited in: PubMed; 17395089.

Ma'ayan A, Jenkins SL, Goldfarb J, Iyengar R. Network Analysis of FDA Approved Drugs and their Targets. Mount Sinai Journal of Medicine. 2007;74:27-32. Cited in: PubMed; PMCID: PMC2561141.

Miller-Jensen K, Janes KA, Brugge JS, Lauffenburger DA. Common effector processing mediates cell-specific responses to stimuli. Nature. 2007;448(7153):604-608. Cited in: PubMed; PMID: 17637676.

Mould DR, Denman NG, Duffull S. Using disease progression models as a tool to detect drug effect. Clin Pharmacol Ther. 2007;82:81-86. Cited in: PubMed; PMID: 17507925.

Natarajan M, Lin KM, Hsueh RC, Sternweis PC, Ranganathan R. 2006. A global analysis of cross-talk in mammalian cellular signalling network. Nat Cell Biol. 2006; 8:571-580. Cited in: PubMed; PMID: 16699502.

Nature News [Published Online]. Drug firm turns spotlight on basic systems biology. 2008 May 7. 453(145), doi:10.1038/453145c. Available from: http://www.nature.com/news/2008/080507/full/453145c.html

Neely M, Jelliffe R. Practical Therapeutic Drug Management in HIV-Infected Patients: Use of Population Pharmacokinetic Models Supplemented by Individualized Bayesian Dose Optimization. J Clin Pharmacol. 2008;48:1081-1091. Cited in: PubMed; PMID: 18635757.

Nestorov I. Whole-body physiologically based pharmacokinetic models. Expert Opin Drug Metab Toxicol. 2007 Apr;3(2):235-49. Cited in: PubMed; PMID: 17428153.

Noble D. Computational models of the heart and their use in assessing the actions of drugs. J Pharmacol Sci. 2008;107:107-17. Cited in: PubMed; PMID: 18566519.

Peterson MC, Riggs MM. Calcium Homeostasis and Bone Remodeling: Development of an Integrated Model for Evaluation and Simulation of Therapeutic Responses to Bone-Related Therapies. The Population Approach Group in Europe [Published Online]. 2007;Abstract1218;16p. Available from: http://www.page-meeting.org/?abstract=1218

Powell JR, Gobburu JVS. Pharmacometrics at FDA: Evolution and impact on decisions. Clin Pharmacol Ther. 2007;82:97-102. Cited in: PubMed; PMID: 175385553.

Riggs M, Peterson M. Development and Utilization of Disease Progression Models Roundtable; Development of a Mechanistic Model of Bone Homeostasis. AAPS National Biotechnology Conference. 2008 Jun 22-25; Toronto, Ontario, Canada. Available from: http://mediaserver.aapspharmaceutica.com/nbc/NBC08/Tuesday/713/Riggs.pdf

Shukla AK, Violin JD, Whalen EJ, Gesty-Palmer D, Shenoy SK, Lefkowitz RJ. Distinct conformational changes in beta-arrestin report biased agonism at seven-transmembrane receptors. Proc Natl Acad Sci. 2008 Jul 22;105(29):9988-93. Cited in: PubMed; PMCID: PMC2481318.

Theil FP, Guentert TW, Haddad S, Poulin P. Utility of physiologically based pharmacokinetic models to drug development and rational drug discovery candidate selection. Toxicol Lett. 2003 Feb 18;138(1-2):29-49. Cited in: PubMed; PMID: 12559691.

Trinh CT, Unrean P, Srienc F. Minimal Escherichia coli Cell for the Most Efficient Production of Ethanol from Hexoses and Pentoses. Applied and Environmental Microbiology. 2008 June:3634-43. Cited in: PubMed; PMCID: PMC2446564.Van Riel NA. Dynamic modelling and analysis of biochemical networks: mechanism-based models and model-based experiments. Brief Bioinform. 2006;7:364-374. Cited in: PubMed; PMID: 17107967.

Urban JD, Clarke WP, von Zastrow M, Nichols DE, Kobilka B, Weinstein H, Javitch JA, Roth BL, Christopoulos A, Sexton PM, Miller KJ, Spedding M, Mailman RB. Functional Selectivity and Classical Concepts of Quantitative Pharmacology. Journal of Pharmacology and Experimental Therapeutics. 2007;320:1-13. Cited in: PubMed; PMID: 16803859.

Vodovotz Y, Csete M, Bartels J, Chang S, An G. Translational Systems Biology of Inflammation, PLoS Computational Biology. 2008 April;4(4):1. Cited in: PubMed; PMCID: PMC2329781.

Wang Y, Bhattaram AV, Jadhav PR, Lesko LJ, Madabushi R, Powell JR, Qiu W, Sun H, Yim DS, Zheng JJ, Gobburu JV. Leveraging prior quantitative knowledge to guide drug development decisions and regulatory science recommendations: impact of FDA pharmacometrics during 2004-2006. J Clin Pharmacol. 2008 Feb;48(2):146-56. Cited in: PubMed; PMID: 18199891.

Wang Z, Zhang L, Sagotsky J, Deisboeck TS. Simulating non-small cell lung cancer with a multiscale agent-based model. Theor Biol Med Model. 2007;4(1):50. Cited in: PubMed; PMCID: PMC2259313.

Wang Z, Birch CM, Deisboeck TS. Cross-scale sensitivity analysis of a non-small cell lung cancer model: Linking molecular signaling properties to cellular behavior. Biosystems. 2008;92(3):249-258. Cited in: PubMed; PMCID: PMC2430419.

Weiner D. Using technology to get more benefit from kinetic and dynamic data. BioITWorld. 2008 Jun 23.

Wingreen N, Botstein D. Back to the future: education for systems-level biologists. Nature Reviews/Molecular Cell Biology. 2006;7:829-832. Cited in: PubMed; PMCID: PMC1950154.

Xu JJ, Hendriks BS., Zhao J, de Graaf D. Multiple effects of acetaminophen and p38 inhibitors: towards pathway toxicology. FEBS Lett. 2008;582:1276-82. Cited in: PubMed; PMID: 18282474.

Yan L, Ropella GE, Park S, Roberts MS, Hunt CA. Modeling and simulation of hepatic drug disposition using a physiologically based, multi-agent in silico liver. Pharm Res. 2008;25(5):1023-36. Cited in: PubMed; PMID: 18044012.

Yan L, Sheihk-Bahaei S, Park S, Ropella GEP, Hunt CA. Predictions of hepatic disposition properties using a mechanistically realistic, physiologically based model. Drug Metab Dispos. 2008;36:759-68. Cited in: PubMed; PMID: 18227144.

Zhang L, Sinha V, Forgue ST, Callies S, Ni L., Peck R, Allerheiligen SR. Model-based drug development: the road to quantitative pharmacology. J Pharmacokinet Pharmacodyn. 2006;33:369-393. Cited in: PubMed; PMID: 16770528.

Zhang L, Athale CA, Deisboeck TS. Development of a three-dimensional multiscale agent-based tumor model: simulating gene-protein interaction profiles, cell phenotypes multicellular patterns in brain cancer. J Theor Biol. 2007;244(1):96-107. Cited in: PubMed; PMID: 16949103.

Zhang L, Wang Z, Sagotsky JA, Deisboeck TS. Multiscale agent-based cancer modeling. J. Math Biology, 2009; 58(4-5):545-559. Cited in: PubMed; PMID: 18787828.

Toolkits and Websites

DEVS modeling framework. Available from: http://en.wikipedia.org/wiki/DEVS

Ptolemy II modeling framework and toolkit. Available from: http://ptolemy.eecs.berkeley.edu/ptolemyII/index.htm

Modeling Works from the Complex Biosystems Modeling Laboratory.

Center for the development of the virtual tumor.