Simbios: Bringing Biomedical Simulation to Your Fingertips

How Simbios' state-of-the-art software tools are contributing to high-impact biomedical research

Simbios began with a simple idea: that physics-based simulation of biological structures at all scales could benefit from a
unified tool-building effort.

 

At the same time, the thinking went, the task of assembling universal tools had to be driven by real research needs—the so-called Driving Biological Projects (DBPs). The center’s initial DBPs therefore involved multiple scales including molecular dynamics (protein-folding, RNA structure prediction, myosin dynamics), neuromuscular simulation, and cardiovascular simulation.

 

Five years ago, Simbios started from scratch trying to meet its ambitious goals. And it has built remarkable momentum in that time. In its first five years, this National Center for Biomedical Computing has assembled a toolkit (SimTK), pieces of which lie within a variety of Simbios applications that address real biomedical research problems at a range of scales.

 

“Ultimately, the hope is that Simbios’ toolkit [SimTK] will provide the underlying—and mostly invisible—computational foundation for a whole range of simulation tools developed both here at Simbios and elsewhere,” says Russ Altman, MD, PhD, and co-PI of Simbios.

 

Getting Simbios’ ideas and tools accepted and used in the wider community is, of course, a major challenge. But by producing high quality, useful tools, making them free and available and providing training and assistance, the center’s tools are finding a welcoming audience.

 

In this feature, you’ll read about how three of the center’s DBPs—neuromuscular, protein-folding, and cardiovascular—have introduced state-of-the-art software tools that are already directly contributing to high-impact biomedical research. You’ll also learn about the wider impact that Simbios is having, based on interviews of thirteen people who are not part of Simbios but who are actively using the center’s tools to advance their own research goals.

 

Scott Delp, PhD, professor of bioengineering at Stanford University, is co-PI of Simbios and leads the OpenSim development team.Simulating human movement is a challenging and complex task, and yet it is essential for understanding how the nervous system coordinates normal movement and how to improve treatments for movement disorders, like cerebral palsy. When Simbios was created, there was no common, open-source platform available for the neuromuscular simulation community to use. In the last two years, Simbios has introduced the OpenSim environment for this purpose. OpenSim is built on the Simbios core software toolkit, called SimTK, including the Simbody multibody dynamics code that Simbios created and released. The OpenSim development team has provided workshops for new users at international conferences. Users provide feedback, develop new features, and perform a wide variety of scientific studies.

 

OpenSim: Neuromuscular Simulation At Your Fingertips

 

When muscles of the arms or legs contract, they tug on tendons and move the underlying bones in ways that can be predicted using the laws of physics. At Simbios, researchers seek to understand the biomechanics of these movements and how modifying them might help treat people with movement disorders such as cerebral palsy, athletic injury or post-stroke limitations.

To achieve these goals, Simbios’ co-PI Scott Delp, PhD, professor of bioengineering at Stanford University, led the effort to create OpenSim, a freely available tool for creating detailed three-dimensional dynamic simulations of human movement.

 

The Tool: OpenSim

OpenSim1 was released to the research community in 2007 (see: Biomedical Computation Review, Fall 2007, p.32), and is now widely used among biomechanics researchers worldwide. OpenSim speeds up the time it takes to do neuromuscular simulations by two to three orders of magnitude, making it possible to run simulations of many subjects performing a task and to do statistical analyses of these simulations. With the upcoming release of OpenSim 2.0 (in October 2009), OpenSim promises to become even more beneficial to researchers in the field, Delp says, offering the capacity to handle constraints (needed for upper-body simulations), simulate contact forces between two bones (useful for the study of osteoarthritis, for example) or with the ground (essential to see the effect of modifying gait). Users will also be able to set their own optimization parameters (giving researchers more flexibility).

 

Courtesy of Katherine Steele and Samuel HammerThe foundation beneath OpenSim has also been strengthened since the initial release. In 2007, OpenSim used only bits and pieces of Simbios’ simulation toolkit SimTK (which includes Simbody, Lapack, Simmatrix, and Simmath). Now that the toolkit is fully developed, OpenSim is built entirely on SimTK, which “provides a very fast dynamics engine to simulate contact between bodies and to do control of multibody dynamics simulations that we haven’t been able to do in the past,” Delp says. “Now, anytime we add a new capability to SimTK, we automatically get it in OpenSim.”

 

The SimTK capabilities can be accessed via OpenSim’s intuitive graphical user interface, allowing users to easily create, animate, and analyze simulations.

 

In the two years since its release, OpenSim has been widely adopted, with more than 3000 users around the world, Delp says. “The field of biomechanics is not that large, and so to have that large number of users is a great success.”

 

The Biology

Using OpenSim, Delp’s team is studying normal and pathological movement, and designing individual treatment plans for those with movement disorders.

 

During walking, muscles must hold the body up against gravity as well as propel the body forward. To quantify the contributions of different muscles to these two tasks at a range of walking speeds, May Liu, PhD, a former student in Delp’s lab created and analyzed 32 walking simulations (eight healthy subjects at four walking speeds). The results were published in the Journal of Biomechanics in November 2008.

 

Among other insights, the work revealed that when people walk slowly (as many with impairments do), they walk with their knees straight, similar to passive dynamic robots. The aligned skeleton holds the body’s weight up, so the muscles don’t have to do this work. But, at faster speeds, people flex their knees to give more shock absorption, and the muscles have to work harder to provide support.

 

The breakthrough in this paper was the number of different simulations run, Delp says. “In the past, generating just a single three-dimensional dynamic simulation of gait was a heroic effort,” he says. “Only one would be developed over a five-year period and then people would analyze that simulation of that one subject.” But with OpenSim, it’s possible to generate many simulations in a few weeks. So, rather than making inferences from just one example, investigators can now do statistics on a sample of simulations. The 32 simulations from this paper are publicly available ( https://simtk.org/home/mspeedwalksims ), so other investigators can easily download and analyze them.

 

Using OpenSim, Delp’s team has also run extensive simulations to study how cerebral palsy affects movement. Patients with cerebral palsy often walk in a crouched position, with an exaggerated bending of the knee. Over time, this condition progressively worsens, leading to joint degeneration and walking difficulties. To study the effect of crouched gait on the muscles of the knees and hip, Delp’s team built simulations based on data from 316 subjects. The work, published last year in the Journal of Biomechanics, found that when people walk with a crouched gait, their muscles have to work harder to fight gravity, because the bent skeleton does not provide this support. At the same time, the muscles lose their ability to straighten the knee and hip. “So it’s kind of a double whammy: the effect of gravity is enhanced and the capacity of your muscles to counteract that is diminished,” Delp says. These effects lead to a worsening of crouch, which leads to more loading on the muscles and further reductions in the muscles’ abilities—and this explains why crouched gait is a downward spiral.

 

Fortunately, it is possible to correct crouch gait—by strengthening specific muscles or through surgery—and reverse it before it progresses. Delp’s team is using simulations of individual patients to predict which corrections will fix crouch in particular patients, and the simulation results are already providing improved understanding of crouch gait and guidelines for treatment. In one project, Delp’s group used OpenSim to simulate dynamics of 127 individual subjects. “That was stunning,” Delp says, "and would not have been possible without advanced simulation software."

 

The Community

Traditionally, the biomechanics simulation community has been fragmented due to the lack of common tools. OpenSim provides a shared platform that can bring the community together. With OpenSim, labs can reproduce, check, and build on each others’ work, Delp says. Users are encouraged to post the simulations they create with OpenSim on the Simtk.org website for anybody to access, and neuromuscular models to use within OpenSim are also shared through Simtk.org. Researchers can also improve OpenSim itself by contributing plug-ins that add functionality to the software.

 

To bring OpenSim to the community, Simbios has held eight workshops and tutorials on OpenSim at Stanford and at conferences around the world; and a ninth workshop is scheduled for October, at which time, OpenSim 2.0—with much greater functionality— will be released. “We have trained over 100 biomechanics researchers during intensive, multi-day workshops at Stanford and have introduced at least three times this many to the software at various conferences,” Delp says.

 

OpenSim has already been downloaded over 5500 times. Researchers are currently using OpenSim to study a wide variety of biomedical problems including joint forces in individuals who are susceptible to osteoarthritis; movement dynamics in individuals with stroke; the movements of the upper limbs in individuals who have a spinal cord injury; and athletic performance and injuries. Besides its use in research, it’s also being widely used in education. “Students in biomechanics labs around the world are using OpenSim as part of their courses,” Delp says.

 

Delp concludes: “What’s become apparent is that because we’ve built OpenSim to be a general biomechanical simulation package, it’s not just serving the specific DBP at Stanford but is now enabling a much broader research community.”

 

Vijay Pande, PhD, associate professor of chemistry at Stanford University, is lead researcher on the Simbios protein-folding DBP.

Protein-folding involves molecular dynamics which uses the rules of physics to simulate the motion and dynamics of proteins. The physics of these molecules requires that small time steps (on the order of femtoseconds, 10-15 seconds) be taken during a simulation in order to guarantee the accuracy of the computed forces and accelerations. However, interesting biology occurs on the timescale of seconds to hours. Simbios has therefore implemented Open Molecular Modeling (OpenMM) to enable the use of graphical processing units (GPUs) to provide 100- to 1000-fold speedups. Simbios is now aggressively disseminating this technology to the biomedical research community, with collaborators at Notre Dame, the University of Pittsburgh, California Institute of Technology, the University of Illinois, the University of Stockholm, in Sweden and elsewhere to help test and use this technology.

 

OpenMM: GPU Acceleration At Your Fingertips

The great limiting factor in simulating molecular dynamics is the speed of computation, says Vijay Pande, PhD, associate professor of chemistry at Stanford University and lead researcher on the Simbios protein-folding DBP. Since the launch of Simbios, Pande’s team has made major advances in the speed of molecular dynamics simulations and is bringing these speed-ups to the community. The team is also at the forefront of a paradigm shift in molecular dynamics, moving away from simulating single long trajectories of molecular events to instead simulating thousands of shorter trajectories. “It’s really about going from something that’s anecdotal—kind of cool, but anecdotal—to something that’s real, statistical, quantitative, and meaningful,” Pande says.

 

The Tools

Last year Simbios released OpenMM2 (Open Molecular Mechanics), an open-source, extensible library that brings GPU capabilities to molecular dynamics software (see: Biomedical Computation Review, Summer 2008, p.28). GPUs, or graphical processing units, are up to two orders of magnitude faster than traditional CPUs (central processing units), but more difficult to program. OpenMM developers (including developers from the GPU manufacturers) did the dirty work—programming core molecular dynamics algorithms across multipGetting Together. Alzheimer’s disease develops when amyloid beta aggregates into neurotoxic clumps (oligomers) and then into plaques. Pande’s team simulated oligomer formation at the all-atom level and analyzed the results using MSMBuilder. This figure shows three different ways that four-chain oligomers (tetramers) can form: (A) as a dimer bound to two monomers; (B) as a dimer plus a dimer; and (C) as a trimer plus a monomer. The simulations predicted that the trimer is the most stable species for aggregates of up to four chains. Reprinted with permission from Kelley, N., et al., Simulating oligomerization at experimental concentrations and long timescales: A Markov state model approach, J Chem Phys 129:214707 (2008), American Institute of Physics.le GPU platforms—and have made these algorithms available to everybody. In early tests, OpenMM achieved speedups of as much as 700-fold compared with traditional CPU implementations. Since GPUs are available on desktops and laptops, this means that scientists can now do fast molecular dynamics using a computer they have under their desk, Pande says. The preview version of OpenMM was released in September 2008 and the software has already been downloaded 2000 times.

 

Since the initial release, “we’ve been rapidly adding functionality,” Pande says. One of the most exciting additions, released in May, is the ability to support explicit solvent models (where water is modeled as individual atoms rather than as a continuous fluid). “For many people molecular simulation means explicit solvent,” Pande says. So this advance gives the software “much, much broader applicability,” he says. “It’s turning out to be the fastest single GPU code to do explicit solvent molecular dynamics,” he adds.

 

Pande’s team is also working on new, faster algorithms for molecular simulations that could “be a game-changer,” he says. With collaborator Jesús Izaguirre, PhD, associate professor of computer science and engineering at Notre Dame University, they have developed a new method (normal mode multiple time stepping Langevin dynamics, or NML) that may speed up molecular dynamics simulations in implicit solvent an additional 10- to 50-fold, Pande says. Then it may be possible to do millisecond simulations in just a few months, he says. Considering that current approaches achieve microsecond simulations in that time frame, “the millisecond timescale is something that most people don’t even talk about. But now it would be something where just about anybody could do it on a desktop. So that gets particularly exciting,” he says.

 

Developers have also added a graphical user interface, called OpenMM Zephyr (released in January 2009 and already downloaded 500 times), which allows experimentalists to take advantage of OpenMM, even if they’ve never done molecular simulations before, Pande says. It is designed to be user-friendly and fail-safe, constraining the naïve user to follow good standard practices for molecular dynamics. “They know that whatever they’re doing with this code, it’s been vetted by us,” Pande says.

 

Finally, MSMBuilder3, a tool that’s complementary to OpenMM, was released in April 2009. OpenMM can simulate thousands of trajectories of a molecular event, such as protein folding, sampled across many time points. MSMBuilder then takes these trajectories and analyzes them statistically to build a Markov state model, which consists of a series of states and the transitions between them (see the “Seeing Science” column on the back cover of this issue for more details). Protein folding is not just about the final structure, but about the intermediate structures it assumes along the way. “Characterizing these intermediates is very much what we’re interested in,” Pande says. MSMBuilder builds a step-by-step model of the folding process—identifying the intermediate states and calculating the transition probabilities between them. The software relies on a novel algorithm, developed with collaborators at Stanford, that clusters intermediate structures based both on geometric similarity as well as kinetic proximity—that is, how energetically easy it is to interconvert between two structures.

 

The Biology

Pande’s lab simulated the folding trajectories of the WW domain. Here, each of four folding trajectories proceeds by a distinct mechanism. Reprinted from Biophysical Journal 96(8), Ensign DL, and Pande VS, The Fip35 WW Domain Folds with Structural and Mechanistic Heterogeneity in Molecular Dynamics Simulations, pp L53–L55 (2009) with permission from Elsevier. Though Simbios’ molecular dynamics tools are relatively new, they are already being used to do groundbreaking science at Stanford and beyond. “The protein-folding and RNA-folding DBPs both have been taking advantage of OpenMM, and I think that has been able to accelerate them pretty dramatically,” Pande says.

 

Using OpenMM, Pande’s team simulated the folding of a protein fragment called WW domain, in the most extensive simulation of this folding event to date. Their results, published in the April issue of Biophysical Journal, reveal that protein folding is a surprisingly heterogeneous process.

 

WW domains are the smallest natural beta-sheet structures (35 to 40 amino acids), and have been extensively studied as a model for beta-sheet folding. Previous simulations focused on generating a single long-folding trajectory—including a landmark 10 microsecond trajectory. But Pande’s group instead ran thousands of simulations of shorter trajectories (some as long as several microseconds) totaling more than 2.73 milliseconds (2730 microseconds). Surprisingly, they discovered that folding proceeded through many disparate pathways and resulted in two distinct end products—the expected 3-stranded beta structure as well as a “misthreaded” version.

 

“This heterogeneity could not have been revealed by any single molecular dynamics trajectory,” Pande says. Capturing this heterogeneity is critical for making quantitative comparisons in experiments as well as for “answering the question that people have been asking me for three decades: how do proteins fold?” Pande says. Citing a single trajectory is like answering the question of “how do people fly from San Francisco to New York?” by citing a single United flight, he says. “This is one anecdotal answer, but the truth is much more complicated and much more diverse.”

 

MSMBuilder is also having substantial scientific impact, for instance, in understanding and potentially treating Alzheimer’s disease. Alzheimer’s disease develops when a certain protein (amyloid beta) misfolds, enabling it to aggregate into neurotoxic clumps (oligomers) and then into plaques. Oligomer aggregation takes place on the seconds timescale, so all-atom simulations of this process were previously impossible. But Pande’s team accomplished such simulations using MSMBuilder; the results were published online in the Journal of Chemical Physics in December 2008.

 

From simulations of more than 6000 short trajectories, totaling 100 microseconds, Pande’s team detailed the structures of 14 intermediate states that may occur during aggregation, as well as the transition probabilities between them. They also predicted that a particular mutation (a glycine-to-proline substitution at position 37) would arrest amyloid beta in an early, non-toxic state; they are currently testing this prediction experimentally. “Such mutants could be useful therapeutically,” Pande says. “People have come up with different schemes for inhibiting amyloid beta toxicity, but this would be a very novel one and we hope that it could go pretty far.”

 

The Community

A major goal of OpenMM (and related software) is to unite the molecular dynamics community, Pande says. OpenMM can bring GPU capabilities to existing molecular dynamics packages. It has already been hooked into GROMACS and ProtoMol, and Pande hopes to extend this to others, such as CHARMM, Amber, and NAMD. “I think it could speed up everybody’s work pretty dramatically,” he says.

 

OpenMM provides a shared interface that can foster collaboration. Because it is completely open, developers can contribute back to OpenMM; and the entire community then benefits from the added functionalities. “If enough people did that, it would make everybody more than the sum of the parts, which would be pretty exciting,” says Pande.

 

To help bring the tools to the community, Pande’s group held free workshops over several days in February and June of this year, attended by 60 scientists and developers from 23 institutions from both academia and industry. The sessions included hands-on tutorials on the use of OpenMM, MSMBuilder, and OpenMM Zephyr, as well as time for developers to work with the OpenMM team on integrating OpenMM into their existing molecular dynamics codes. Additional workshops will be held next year.

 

“There are a lot of interesting things to think about for the next five years,” Pande concludes. “But even now, we’ve gotten to a point where we’ve got very powerful tools. And I’d love to see what people will do with them.”

 

Charles A. Taylor, PhD, associate professor of bioengineering at Stanford University, is PI for the cardiovascular dynamics project within Simbios.Cardiovascular disease is a primary source of morbidity and mortality in the United States and the world.  Fundamental to understanding the mechanisms of this disease, and for formulating strategies for treatment, is the ability to simulate both the normal cardiovascular system as well as the system in disease. SimVascular is an environment for fluid dynamic simulations that provides advanced modeling capabilities and allows users to move from clinical images (CT and MRI) of the vascular system to 3-D static models and then on to dynamic models of circulation. These models are notable because they are among the first to produce quantitatively accurate measures of pressure and flow for the human vascular system. SimVascular is now available to scientists and clinicians around the world.

SimVascular: Cardiovascular Simulation At Your Fingertips

When the heart or vascular system becomes diseased, medical imaging provides a window into the problem—but the information is limited. Simbios’ cardiovascular simulation program SimVascular picks up where imaging leaves off. From imaging data, SimVascular reconstructs an accurate three-dimensional model of blood flow through the arteries of individual patients; this model can be used to predict outcomes and virtually test interventions.

 

SimVascular is giving doctors new insights into arteriosclerosis, aneurysms, heart disease, and congenital heart defects; guiding the development of better medical devices; and being used to plan the treatments and surgeries of individual patients, says Charles Taylor, PhD, associate professor of bioengineering at Stanford University and PI for the cardiovascular dynamics project within Simbios.

 

The Tool: SimVascular

SimVascular is a simulation program that combines several state-of-the-art commercial components with open-source code. It is fully parallelized and highly scalable, Taylor says. “The software has been run on computers with tens of thousands of processors and on a laptop,” he says.

 

Virtual Vessels. A patient-specific model from SimVascular is superimposed on the computed tomography (CT) image data for the same patient. This patient has a cerebral aneurysm (shown in blue on the left side of this picture).  Reprinted with permission from: Taylor CA and Figueroa CA. Patient-Specific Modeling of Cardiovascular Mechanics. Annu Rev Biomed Eng 2009; 11:109-134.  SimVascular distinguishes itself from other cardiovascular simulation programs because it accurately models blood pressure as well as blood flow velocity. “That’s very unique—we can model not just the velocity of blood going through the arteries, but the actual realistic pressure wave propagating through the vascular system,” Taylor says. “Blood pressure is obviously quite important, because how much the arteries deform is directly related to blood pressure.”

 

Other cardiovascular simulation programs have ignored the deformability of arteries, treating them as rigid tubes; but SimVascular treats blood vessels realistically—as flexible, dynamic objects that interact with the blood. It is also unique in that it models the microcirculation—the tiny blood vessels at the ends of arteries that cannot be seen through imaging but play a critical role in blood pressure, distribution, and flow. “In order to be able to predict the outcome of an intervention, you have to have a realistic model of the microcirculation,” Taylor says.

 

Since the beta version of SimVascular was released in 2007 (see: Biomedical Computation Review, Spring 2007, p. 25), the code has been “tested and improved,” he says. An updated version was released in summer of 2008 that yields more stable solutions for problems that previously caused the program to crash. “That basically enabled us to solve a much larger number of problems than we could solve before,” Taylor says.

 

He and others are working on several new capabilities that may come online with future releases of SimVascular. For example, the “fluid-solid-growth” model connects SimVascular with software that models the growth and adaptation of blood vessels over time (developed by Jay Humphrey, PhD, professor of biomedical engineering at Texas A&M University). Currently, SimVascular simulates at most 50 cardiac cycles (about 50 seconds). The fluid-solid-growth model can simulate events over longer time periods—such as the growth of an aneurysm over years, Taylor says.

 

Another major advance that Taylor’s group is working on—which he is calling “image-based fluid-solid interaction”—enables SimVascular to build models from four-dimensional imaging data, such as time-resolved CT or MRI. Unlike three-dimensional imaging data, which is averaged over the cardiac cycle, four-dimensional data gives the instant-by-instant changes in geometry and would yield more realistic simulations, Taylor says.

 

Finally, the commercial components of SimVascular have posed some difficulties for users. “It's a challenge to compile right now because of all of those parts and issues of proprietary software,” says Alison Marsden, PhD, an assistant professor of mechanical and aerospace engineering at the University of California, San Diego. Users also have to pay license fees for the commercial components or search for suitable open source alternatives, which aren’t readily available. So, Mardsen and others are hoping to develop open-source alternatives and contribute them back to the code.

 

The Biology

Among other applications, Taylor’s team is using SimVascular to study interventions for abdominal aortic aneurysm, a potentially deadly condition in which the abdominal aorta weakens, bulges, and might eventually rupture.

 

The condition is usually treated with a stent graft: doctors insert a fabric-covered metal tube into the diseased vessel, and blood flow is diverted through the tube, removing pressure from the aneurysm. Unfortunately, some stent grafts eventually dislodge and move over time, letting blood back into the aneurysm and putting the patient at risk for rupture.

 

Doctors have always assumed that the grafts move parallel with the blood vessel—flowing downstream. But in two papers in the June 2009 issue of the Journal of Endovascular Therapy, Alberto Figueroa (a research associate in the bioengineering department at Stanford), Christopher Zarins (a professor of surgery at Stanford) and Taylor demonstrated that the forces acting on stent grafts are in fact pushing the grafts sideways (perpendicular to blood flow) in the majority of cases. These simulation results obtained with Simvascular help to explain the observed lateral movement of stent grafts recently described by Zarins.

 

Using SimVascular, Taylor’s team built three-dimensional models of the abdominal aorta based on imaging data from individual patients with aneurysms. Then they placed virtual stent grafts in these models and calculated the magnitude and direction of the forces acting on the grafts. Surprisingly, the majority of the forces were acting perpendicular, rather than parallel, to the graft. The more curved the artery, the bigger the sideways force—and “most aortic aneurysms are very tortuous,” Taylor says. The finding has huge consequences for device design, Taylor says. “So that’s had a lot of repercussions in the medical device industry.” He predicts that the two papers will be highly cited.

 

Taylor’s team is also involved in a rare pairing of computational simulation with a clinical trial. SimVascular is playing an integral part in a large, ongoing trial to test the potential for exercise to slow the growth of abdominal aortic aneurysms. The clinical results from the three-year trial are not due out until next year, but the simulation results are already yielding insight into the hemodynamic effects of exercise.

 

Using imaging data from patients in the trial, Taylor’s team simulated the effect of exercise on aneurysms with a variety of different geometries—and found that they all derived benefits. “Light exercise was sufficient to eliminate areas of chronic blood flow stagnation, which are hypothesized to lead to the progression of aneurysms,” Taylor says. A paper describing the results is under review.

 

The Community

SimVascular is impacting the cardiovascular research community in a variety of different spheres. To teach the community about SimVascular, Stanford has hosted two short courses (in 2007 and 2008), which were attended by participants from both academia and industry. Taylor has also co-founded Cardiovascular Simulation, Inc., a company that will use SimVascular in-house. The company will partner with pharmaceutical companies, medical device companies, and doctors to help bring the capabilities of SimVascular—such as planning better surgeries and designing better devices—outside of academia.

 

Taylor and Humphrey are also leading the vascular part of the Physiome project, which is an international effort to provide an integrated framework for modeling human physiology at the cellular, tissue, organ, and total-body levels. SimVascular will play a vital part in this effort. “The Physiome is a global project, with global outreach. That’s been a major focus,” Taylor says.

 

Simbios: Packaging Research Tools for Your Fingertips

Simbios does first-rate science, says Jeanette Schmidt, PhD, executive director of Simbios. But what distinguishes Simbios from other research projects at large research universities is the dissemination effort. “Research universities don’t often do such a good job of disseminating software tools,” she says. “But at Simbios, we spend a lot of time packaging these things up so that other people can actually use them.”

 

In its first five years, in addition to releasing OpenSim, OpenMM and SimVascular, as described above, Simbios released tools (and produced significant research) related to two other major efforts—the RNA Folding DBP and the Myosin Dynamics DBP.
In providing these tools, Simbios went the extra mile to make them user-friendly. It also fulfilled its aim to provide simulation tools across a range of scales, Schmidt says, with a shared underlying toolkit (SimTK) used at both the neuromuscular and the molecular scales.

 

With this foundation in place, Schmidt says Simbios is well-primed for the next five years. “I really think we have built the right tools,” she says. “But there’s still significant refinement and additional functionality needed in order to make them applicable to more areas.”

 

The plan, then, is to emphasize more sophisticated use of the tools Simbios has already built—to exploit what has already been done, improve it and expand it into even more research areas, Schmidt says. “The opportunity is really there for the fruits to be reaped.”

 

 

 

 

 

 

The featured Simbios community members and the tools they use

Jill Higginson, PhD: http://biomedicalcomputationreview.org/content/opensim-user-profile-jill...

Silvia Blemker, PhD: http://biomedicalcomputationreview.org/content/opensim-user-profile-silv...
Katherine Holzbaur, PhD: 3) http://biomedicalcomputationreview.org/content/opensim-user-profile-kath...

BJ Fregly PhD http://biomedicalcomputationreview.org/content/opensim-user-profile-bj-f...

 

Erik Lindahl, PhD: http://biomedicalcomputationreview.org/content/openmm-user-profile-erik-...

Jesus Izaguirre, PhD: http://biomedicalcomputationreview.org/content/openmm-user-profile-jesus...
Kim Branson, PhD http://biomedicalcomputationreview.org/content/openmm-user-profile-kim-b...
 

 

Jay Humphrey, PhD: http://biomedicalcomputationreview.org/content/simvascular-user-profile-...
Alison Marsden PhD: 12) http://biomedicalcomputationreview.org/content/simvascular-user-profile-...

 

Rick Russell, PhD: http://biomedicalcomputationreview.org/content/rna-builder-user-profile-...
Li Niu, PhD: http://biomedicalcomputationreview.org/content/nast-user-profile-li-niu-phd
Jung Chi Liao: http://biomedicalcomputationreview.org/content/allopathfinder-user-profi...
Alain Laederach, PhD: http://biomedicalcomputationreview.org/content/simtk-user-profile-alain-...

 

References

1. Friedrichs S, Eastman P, Vaidyanathan V, Houston M, LeGrand S, Beberg AL, Ensign DL, Bruns CM, Pande VS. Accelerating Molecular Dynamic Simulation on Graphics Processing Units J Comp Chem 2009;30:864-872.
2. Bowman GR, Huang X, Pande VS. Using generalized ensemble simulations and Markov state models to identify conformational states. Methods (article in press, 2009).
3. Delp SL, Anderson FC, Arnold AS, Loan P, Habib A, John CT, Guendelman E, Thelen DG. OpenSim: Open-Source Software to Create and Analyze Dynamic Simulations of Movement. IEEE Transactions on Biomedical Engineering 2007; 54: 1940-1950.



All submitted comments are reviewed, so it may be a few days before your comment appears on the site.

Post new comment

The content of this field is kept private and will not be shown publicly.
CAPTCHA
All submitted comments are reviewed, so it may be a few days before your comment appears on the site. This question is for testing whether you are a human visitor and to prevent automated spam submissions.
Image CAPTCHA
Enter the characters shown in the image.