The New Model
What the 2013 Nobel Prize in Chemistry says about predictive sciences
The 2013 Nobel Prize in Chemistry, announced on October 9, 2013, was awarded to Martin Karplus (Harvard), Michael Levitt (Stanford) and Arieh Warshel (University of Southern California), “for the development of multiscale models for complex chemical systems.” It recognized their work in bringing together classic Newtonian molecular mechanics and quantum mechanics methods to help scientists model large complex chemical systems and reactions. Given the magnitude of the achievement, this recognition was widely welcomed and, arguably, overdue.
What fascinated me was the associated information provided by the Nobel organizing committee, in particular, this short statement that emphasizes the wider impact: “Today chemists experiment just as much on their computers as they do in their labs.” The Nobel chemistry committee is highlighting the value that predictive science adds to our understanding of how molecules interact in chemistry and biology. It is appropriate to do so: the predictive multiscale models and methods developed by Karplus, Levitt, Warshel and others have fundamentally changed the way scientists innovate in bringing new drugs to market. For instance, many pharmaceutical modeling tools are based on CHARMM (chemistry at Harvard macromolecular mechanics) – a molecular dynamics simulation and analysis program that was first developed by Karplus’ group decades ago. My company’s product, the Accelrys Discovery Studio, contains the only commercial version of this widely used program.
It’s not so long ago that we were using Orbit kits or the metal Dreiding kits to build molecular models. In the ‘good old days’, these tools helped us to understand the underlying physics of molecular structures but couldn’t tell you anything about how molecules moved, what forces of strain they had within them or, crucially, how they interacted with other molecules. Fortunately, while we were playing with sticks and balls, mathematical models of molecular systems quietly but steadily evolved, progressing from purely theoretical tools that were used by just a few highly specialized computational scientists into robust and broadly accurate predictive tools that are readily applicable to discovery research applications. These have transformed the way that many areas of scientific research are now conducted. A much broader population of scientists can now routinely access and use them; indeed, such tools are now starting to make an appearance on smart phones and tablet devices.
To demonstrate the paradigm shift from theoretical tools into everyday research aids, cast your mind back to the heady days when computational chemists ran simulations on huge (and usually very hot) computing platforms. These were typically power-hungry, expensive machines that were hidden away in broom closets and far removed from “real chemists.” Today, if you walk around any lab you will see those same chemists actively collaborating with their project team counterparts using new generations of the same tools. They might, for example, be calculating the properties of a macromolecule or predicting how a compound may bind to a protein target. Today, we consume predictive science directly; it is no longer segregated from the discovery process.
An area where this is of particular value is biotherapeutics. Many of modern medicine’s most valuable tools for treating and preventing illness are biologics and topics such as the development of novel antibody technologies have become key research pursuits. However, biotherapeutics also present unique challenges. Their storage and administration typically requires high-concentration solutions with specific biophysical profiles pertaining to solubility, thermal and chemical stability, and low aggregation propensity. Testing for these properties can be time-consuming and expensive. To accelerate development and reduce costs, researchers are increasingly applying predictive methods to identify and optimize the best biologic leads early.
By enhancing direct experimentation, the multiscale modeling techniques developed by this year’s Nobel laureates in Chemistry can make these and many other hitherto unsolvable problems solvable.
Adrian Stevens is the Senior Product Marketing Manager for Discovery Studio, Accelrys’ premier Life sciences modelling and simulation product. He has more than 12 years experience in the practical application of computational chemistry; the vast majority of this time spent directly within the pharmaceutical industry. He received his PhD in Computational Chemistry from the University of Portsmouth in 1996, and subsequently continued in a postdoctoral position, focusing on the novel application of molecular spectra as descriptors to model drug activity. As well as having presented numerous papers at international conferences, Adrian has been actively involved in the computational science community, both in the organisation and support of conferences, such as the original “Cutting Edge Approaches to Drug Design” series, and in the “Young Modellers Forum”.
