Driving Biological Projects

Our criteria for the selection of our Driving Biological Projects (DBPs) are outlined below. Our portfolio of DBPs should individually, or in aggregate, satisfy these criteria. The criteria below drove our choice of our initial DBPs, and will drive our choice of problems throughout the life of the Center.

  • Canonical   The problem should be an archetype of the problems that occur in an entire field of inquiry, so that results are guaranteed to have broad applicability.

  • Cover a range of scales   To exercise our capabilities and identify our shortcomings in multi-scale modeling, it is critical that each DBP have some range of scales at which questions can be posed.

  • Physics-based   The DBP should present a problem that can be addressed by representing and analyzing the geometry and physics of the biological system.

  • Be data rich   Biology is dominated by experimental data. These data provide the real-world constraints that drive and validate models and simulations.

  • World-class, engaged experimentalists  

  • Collectively cover a broad area of biophysical modeling   We cover a broad area of scales, and attempt to create an environment relevant to molecular biologists, cell biologists, neurologists, neuroscientists, and surgeons.

  • Have important implications for disease   Our DBPs are related to disease processes or treatments, to ensure that they contribute to the overall advancement of human health. RNA structure has implications for rheumatologic diseases, protein folding has, among many others, implications for Alzheimer, myosin dynamics is important for understanding myopathies and the generation of motive force throughout all organ systems, neuromuscular dynamics must be understood to better treat movement disorders such as cerebral palsy, stroke and Parkinson's disease, and cardiovascular dynamics has important implications for coronary artery and peripheral vascular disease.

DBPs change over time. As a new DBP is added, funding allocated to a previously active DBP may be reduced or stopped. In the DBP Graduation at Simbios document we describe our plan for continued scientific activity and software support.

Our DBPs are briefly reviewed here:

TYROSINE KINASE DYNAMICS (introduced Fall 2013)

Protein kinases are a family of enzymes that can alternate between active and inactive states in response to specific signals. When they are not regulated, uncontrolled cell proliferation and malignant transformation as observed in numerous cancers can result. This makes kinases an important target for therapeutic intervention. Atomistic molecular dynamics simulations of kinases could capture these conformational transitions and provide new insights into and data for designing inhibitors. Dr. John Kuriyan, professor of molecular and cell biology and of chemistry at U.C. Berkeley and HHMI Investigator, leads this new DBP. Combining the experimental expertise of his lab with the computational techniques developed in the lab of Simbios faculty member Dr. Vijay Pande will drive our biological knowledge and the computational methods in new directions.


Stroke is the leading cause of serious long-term disability in adults worldwide, and walking dysfunction is one of the greatest stroke-related physical limitations. While approximately two-thirds of persons who suffer a stroke regain ambulatory function, their gait is slow, asymmetrical, and metabolically inefficient. Despite recognition of the problem, there is limited evidence that generic rehabilitation methods produce meaningful changes in walking function. Biomechanical simulations have the potential to transform current treatment approaches from a subjective, qualitative process into an objective, quantitative one that would be more effective at improving walking function in persons post-stroke. Dr. BJ Fregly, a professor of Mechanical and Aerospace Engineering at the University of Florida, is the PI of this DBP. He is working closely with Simbios’ OpenSim team, headed by Dr. Scott Delp, to use OpenSim to predict the best achievable gait pattern in patients who have experienced a stroke.


The long-term goal of this project is to develop arm prostheses for amputees that can be directly controlled by the brain. Achieving this goal requires decoding motor intention from recordings of brain activity during complex movement patterns. Towards that end, Dr. Krishna Shenoy and Dr. Michael Black are leading efforts to 1) establish a freely moving animal model to directly measure the context-dependency of motor cortical activity and 2) develop computer-vision algorithms and biomechanical models to automatically determine body and limb orientations during free movement over long periods of time. Leaders in their respective fields of neuroprosthetics and computer vision, Drs. Shenoy and Black are collaborating with computational scientists (Drs. Guibas and Latombe) and biomechanical expert (Dr. Delp) on this project. The integration of neuroscience, computer vision, and biomechanical modeling will enable the unprecedented study of motor control during natural behavior. This new paradigm will greatly enhance neuroscience investigations of motor control, advance neuroengineering studies aimed at designing high-performance neural prostheses, and improve the quality of life for individuals with physical disabilities by restoring lost motor function.

DRUG TARGET DYNAMICS (introduced Fall 2010)

This project is developing physics-based methods to improve drug-docking and the modeling of unexpected drug-target interactions. Specifically, it seeks to understand how physical simulations can improve docking for G-protein coupled receptor proteins (GPCRs), which constitute approximately 50% of all drug targets. While large-scale docking experiments, such as those for GPCRs, typically do not include long time-scale molecular dynamics due to a lack of computational resources and tools, recent Simbios innovations present new research opportunities. Simbios software OpenMM and MSMBuilder enable fast dynamics and clustering, so that target dynamics can be included and the resulting trajectories combined with other chemical and biological information to improve our ability to predict interactions between targets and ligands. Leading computational chemist Dr. Brian Shoichet of U.C. San Francisco heads up the project in collaboration with Simbios researchers (Drs. Pande, Altman, Levitt, Guibas, and Hanrahan).

RNA FOLDING (introduced Fall 2004)

This project represents a macromolecule-scale study of the process of folding. It is ultimately relevant to RNA, DNA and protein, but focuses on RNA as an intriguing experimental system. The folding of RNA is a function of the physical forces that act on it. Dr. Herschlag has assembled a world-class team from the Stanford Departments of Biochemistry, Genetics, Physics, and Applied Physics. Three external scientists (Drs. Pollack, Brenowitz, and Chance) are also participating. The importance of RNA function has recently been magnified with the discovery of RNA inhibition and a large number of microRNA genes. RNA, even more than DNA, seems to implement its functions using complex structural strategies. Although the primary scale of the RNA work is at the atomic level, certain computations are too expensive to perform atomistically; thus, the project requires coarse-grain representations in which, for example, a single base is represented as a ball, or a segment of A-form double helix is represented as a cylinder.

PROTEIN FOLDING (introduced Spring 2008)

This driving biological problem investigates the folding kinetics of proteins and protein-protein complexes. Due to the limitations of both simulation and experiment, an ultimate understanding of protein folding will likely come from a coupled approach of detailed simulations extensively validated and tested by experiment. Dr. Pande is leading this project, which hence naturally has an active collaboration with folding@home. A key aspect of this DBP is the development of OpenMM (Open Molecular Mechanics), an extensible API for molecular mechanics. OpenMM is designed to operate with the tens of different Molecular Dynamics codes with overlapping functionality, each with their own user base, and which currently somewhat results in a fragmented community. OpenMM is designed to take hardware acceleration into account, in particular computations on Graphic Processor Units (GPUs). Hence OpenMM can act as a unifying API the way OpenGL unified the graphics community. The OpenMM API would be used as the backend to existing codes, allowing for all to benefit from hardware acceleration.

MYOSIN DYNAMICS (introduced Fall 2004)

This project represents a scale that is one order of magnitude larger than RNA. Myosin is a large protein, and, in a muscle cell, organizes itself into fibers. Myosin represents the fundamental source of motive force in many living systems, and so its biological importance is high. It is fundamentally a physical problem: how does the cell turn the chemical energy of ATP into movement? Experience with myosin will no doubt improve our ability to address other molecular machines. Dr. Spudich has a distinguished history as a leader in the field of myosin biology. The relevant scales for myosin range from Angstroms to nanometers, as the molecules assemble into larger aggregates. Dr. Spudich's project uses a range of experimental biochemical, molecular biological, biophysical and genetic experimental techniques to approach this problem, so there is ample data, and (like the RNA project) a very real opportunity to collect additional data if it should be required by the modeling and simulation effort.

NEUROMUSCULAR DYNAMICS (introduced Fall 2004)

The range of scales relevant to this problem is impressive: the precise physical properties of a muscle cells at the micron scale, all the way to the macroscopic forces generated by muscles on the scale of centimeters. Although this DBP focuses on walking, the findings will be generalizable to other motor control systems. The modeling of human motion is a biomechanical and physical problem, and Dr. Delp is a pioneer in the development of methods for modeling motor systems. The functional implications of this work are paramount: a primary application of this work is in the planning of interventions to assist patients with abnormal movement dynamics, including children with cerebral palsy, and adults with stoke and Parkinson's disease. Advances in imaging and instrumentation provide rich data sets for building and evaluating neuromuscular models.

CARDIOVASCULAR DYNAMICS (introduced Fall 2005)

This project represents scale from millimeters to meters, and focuses on the dynamics of fluid flow through the branching system of blood vessels in the human cardiovascular system. While focusing particularly on aortic structure and strain, the findings are generalizable to other flow systems. This DBP is important because the physics of fluid flow are markedly different than the physics of multi-body dynamics used at the molecular and neuromuscular level. This DBP is therefore will stress our systems for representation and organization more than others, and guarantee generality. Drs. Zarins and Taylor are leaders in patient-specific modeling of vascular flow, and its use in surgical bypass planning. Thus, the clinical significance is high.

These DBPs provide a challenging set of diverse object types and underlying relevant physics. The research challenges outlined are drawn directly from an analysis of the modeling and simulation challenges required to unify biophysical modeling and simulation. The multiscale nature of each of these DBPs is important. Artificial disciplinary boundaries separating different modeling and simulation communities (e.g., structural biology and mechanical engineering) have limited the creation and application of multiscale simulation techniques in biomedicine. The recognition that these boundaries limit our ability to perform useful simulations makes their removal critical for long-term success. We believe that multiscale capabilities are not a set of features that can simply be added to existing disciplinary packages for simulation. These packages are predicated on fundamental assumptions that make them effective within a certain dynamic range, and ineffective outside. Instead, multiscale capabilities must be built as part of the founding concept of a simulation environment.