CSEM Alumnus Juan M. Bello-Rivas has won the ICES 2017 Outstanding Dissertation Award for advances in the computational simulation of molecular dynamics.
Now a postdoctoral researcher at Princeton University, as a CSEM student at UT Bello-Rivas focused his dissertation on modifying a statistical mechanic theory and algorithm called Milestoning. The algorithm was originally developed by Ron Elber, his advisor in the CSEM program.
The professor of chemistry and biochemistry and director of ICES’ Center for Computational Life Sciences and Biology unveiled Milestoning in 2004 as a way to model dynamic changes that occur over a computationally long time (milliseconds to seconds) within biomolecular systems without losing atomic-level spatial resolution. Because of its capabilities, Milestoning can be used to improve the understanding of processes such as the transmission of a signal into a cell to initiate cellular division, or the ability of an anti-cancer drug to attach to an abnormal protein to thwart rampant cell growth.
Under Elber’s guidance, Bello-Rivas focused his research that he defended in November 2016 on designing algorithms and performing mathematical analyses that allow Milestoning to more accurately simulate the interactions within biomolecular systems. The refinement Bello-Rivas developed over four years, known as Iterative (or Exact) Milestoning, can be applied to a broader set of biomolecular systems, including non-equilibrium systems where an external force is applied. In addition, the approach is scalable to whatever level of accuracy is needed (dependent on computational limits).
“This refinement of Milestoning requires less assumptions about what interactions are occurring in a system, without using methods of approximation,” said Bello-Rivas, whose thesis was chosen for the award by an ICES committee. He is now a postdoctoral researcher at Princeton University’s Program in Applied and Computational Mathematics.
In particular, Iterative Milestoning overcomes the difficulties of doing calculations related to the long time spans that a molecule of interest could spend transitioning between intermediate energy states. Those states often occur before a biomolecule reaches a global energy minimum (the molecule’s most stable position, such as the completely folded state of a protein that helps form a signal binding site (receptor) on a cell surface). Iterative Milestoning does so by compressing the time spent on analyzing intermediate energy states, and addresses the molecule’s overall behavior through a focus on its smaller, more easily analyzable, components.
For instance, rather than studying five different positions that a carbon atom of a protein could occupy in space depending on the protein’s potential shapes, Bello-Rivas confirmed that you could restrict the dynamic simulation to represent only a few of the atom’s potential conformations. Calculations done on the state of the system based on fewer conformations can then be analyzed in a repetitive (iterative) way using parallel computing methods to refine the understanding of the entire protein. That involves identifying the set of probability distributions related to a molecule’s existence in a particular conformation, and then feeding the findings back into additional rounds of analysis in a way that improves the accuracy of the details of the molecule’s potential conformations.
“Juan made an important leap in iterative Milestoning by making it more mathematically rigorous, which is essential for verification of the calculations,” says Elber, the W. A. “Tex” Moncrief Chair in Computational Life Sciences and Biology. “Equally important, he opened the door for Milestoning’s use to improve the understanding of non-equilibrium simulations that are critical for many biological applications such as transport.”
A native of Santiago de Compostela, Bello-Rivas earned a bachelor’s and a master’s degree in mathematics from the Universidad Complutense de Madrid. As part of his dissertation at UT Austin, Bello-Rivas also overcame the challenge of using general software packages to run molecular dynamics simulations with Milestoning when computing kinetic properties of molecular systems. He developed a software package that automates the running of Milestoning simulations while relying on standard molecular dynamics codes to be used on high performance computers such as those at the university’s Texas Advanced Computing Center.
Despite the intensity of his research and course work, Bello-Rivas co-authored three articles while in Austin, including two as lead author. One of those on Exact (or Iterative) Milestoning was named among the most innovative and influential of the year by the Journal of Chemical Physics.
A book chapter and two other Milestoning articles are pending. But that’s not all, as demonstrated by an article collaboration with David Aristoff, an assistant professor of mathematics, at Colorado State University that he participated in while at UT. As Elber notes, "Juan also excelled in bridging different fields, taking physical chemistry theories and bringing in formal mathematics.”
--Written By Barbra A. Rodriguez