Carlos Simmerling Discusses the Importance of Computer Simulations
Fast Moving Front Commentary, May 2011
Article: Comparison of multiple amber force fields and development of improved protein backbone parameters
Authors: Hornak, V;Abel, R;Okur, A;Strockbine, B;Roitberg,
A;Simmerling, C |
Carlos Simmerling talks with ScienceWatch.com and answers a few questions about this month's Fast Moving Fronts paper in the field of Biology & Biochemistry.
Why do you think your paper is highly cited?
The importance of computer simulations keeps growing in a wide variety of areas, including my area of expertise, which is modeling the behavior of molecules important to biology. Experiments are wonderful but can only tell us so much—accurate computer models can provide information that is very difficult to obtain directly with experiments.
The accuracy of these simulations depends on the quality of the "force field" that emulates the underlying physics. These force fields describe how the energy of the molecule responds to various events needed for function, such as when they change their shape, or when they interact with water, with each other or with small molecules such as pharmaceutical compounds. Because of the large size of biomolecules, we need to make many approximations, and the force field only provides an estimate of the true energy, which controls all of the behavior.
In this paper, we reported significant improvements to methods used to evaluate force field accuracy, and used the results to develop a better energy function that is now used by a large number of labs that simulate biomolecular systems. The work was done in partnership with my close friend and longtime collaborator, Adrian Roitberg at the University of Florida.
Does it describe a new discovery, methodology, or synthesis of knowledge?
This image shows a snapshot from a simulation that
revealed for the first time how the AIDS drug target HIV protease opens to
allow a small molecule inhibitor (green) to enter and inactivate it. The
protein is shown in a schematic ribbon representation; the simulations were
done at the atomic level.
The image shows a snapshot from a simulation exploring
how a DNA repair enzyme binds to and bends the DNA duplex in its search for
damage. DNA repair occurs thousands of times a day in every cell and every
organism.
The image shows how simulations can study the folding of
small proteins and gain insight into the underlying landscape (colored by
energy) that controls folding, giving insight into how mutations can cause
misfolding and result in diseases such as Alzheimer’s. Various
structures are drawn near the bumps in the landscape as the flexible
protein follows the path to adopting the functional native
structure.
It's hard to put it into these categories. We know that our models always involve compromises and have limitations. Biomolecules are just too big to treat using methods that we know are more accurate, like quantum mechanics. Our advance was based largely on much earlier work by pioneering researchers like Peter Kollman, who worked on small fragments of biomolecules where the accurate quantum mechanics calculations were tractable. They used the results to train simple force fields, which could then be used for much larger biomolecules.
My team built on that approach, but we recognized that growth in computer power now enabled us to get quantum mechanics data for much bigger fragments—ones that actually looked more like real biomolecules. As part of training our force field, we also did a more extensive comparison to experimental data than had been done before. We expected that a better description of the physics for the small fragments would lead to better simulations on big biomolecules as well. Although the changes were relatively simple, the results from simulations using the new force field turned out to be significantly better than with previous models. We are pleased that many other labs have also done careful testing of our force field and confirmed that it gives much better agreement with experimental data.
Would you summarize the significance of your paper in layman's terms?
Scientists want to be able to use computers to do things like design drugs that are more effective, or to understand why genetic variations or DNA damage can give rise to disease, or to improve our ability to obtain fuels from renewable resources. For this to work, we need to improve the quality of the computer models so that our molecular simulations accurately reflect what will happen in the real world.
In our paper, we developed an improved model that can give results that are significantly better than what researchers could obtain before. This model has been widely adopted by other labs. I'm very happy that my team's work could enable such a wide variety of interesting studies.
How did you become involved in this research, and how would you describe the particular challenges, setbacks, and successes that you've encountered along the way?
Ever since I was a child I wanted to be a chemist and make new molecules with fun and useful properties. During college, though, I found that the lab environment wasn't right for me and I got very discouraged. Later, I was lucky enough to find a group using computers instead of test tubes, and it transformed my vision for my career. Since the molecular simulation field is not yet mature, I spend about half of my research effort trying to improve our models and computer software, and the other half using these tools to study biological problems.
My team and I are always comparing to experiments to find out when and where our models fail, and it's always frustrating to be reminded of our limitations. We try to use these setbacks to learn how to improve our methods. For example, when I first got my job as an assistant professor I worked on simulating protein folding, and failed because the models just weren't good enough. We didn't give up, and our analysis of patterns in the errors led directly to the design of the new force field that is the subject of this paper. It led me to being able to do much better scientific research than if I had not tried to learn from my failures.
Where do you see your research leading in the future?
In the short term, we're always improving the models, and we recently completed a new force field that is even better than what we reported in this paper. Farther off, I think the future for computational modeling is very exciting in many areas, not just the physical sciences. With the growth in supercomputer power, teams in my field are starting to be able to not just reproduce things we already knew from experiments, but to gain insight into underlying causes for the experimental observations, and even to help design new experiments and new molecular systems.
We'll keep getting better at connecting the huge gap in scale between the molecules and the whole organism, but we have a very long way to go before we have real physical models of how differences of a few atoms in the DNA code of a genome can influence things like changes in behavior or susceptibility to disease. Faster computers, more efficient algorithms, and better physics models are being developed by many labs, and our field makes big steps forward every year. In my opinion it's a thrilling time to be a computational scientist.
Do you foresee any social or political implications for your research?
Many of the most challenging problems faced by the world today are really issues of chemistry—access to clean water, effectiveness and cost of medication, quality and quantity of food, and low-cost renewable energy. Researchers around the world are studying the behavior of molecules to try to solve these problems. I believe that computer models that are good enough to help understand and guide experiments could be a key aspect of success, and along with many other labs, I'm trying to help make that happen.
Carlos Simmerling
Professor, Department of Chemistry
and
Associate Director, Laufer Center for Physical and Quantitative
Biology
SUNY Stony Brook
Stony Brook, NY, USA
KEYWORDS: TRIALANINE, DIHEDREAL PARAMETERS, MOLECULAR DYNAMICS, MOLECULAR MECHANISMS, DECOY ANALYSIS, NMR ORDER PARAMETERS, ALPHA HELIX, 2-DIMENSIONAL VIBRATIONAL SPECTROSCOPY, FREE-ENERGY CALCULATIONS, SEQUENCE CULLING SERVER, MAGNETIC RESONANCE DATA, N-15 NMR RELAXATION, EGG-WHITE LYSOZYME, POLYPROLINE II, AQUEOUS SOLUTION, DISORDERED CONFORMATIONS.