2016 ICES Postdoctoral Fellow Ernesto Lima's algorithms for cancer tumor models (pictured) have helped evolve the field's work from sophisticated representations of cancer’s theoretical behavior to actual predictive tools.
For the past five years, Ernesto Lima has been making computational models of tumor growth. The models use complex mathematics to simulate the eight hallmarks of cancer behavior—abilities all cancers share, such as unrestricted cell division and immune system evasion. Yet the guiding philosophy of the work is simplicity: the best model for simulating a tumor’s behavior is the one that uses the lowest number of parameters from the available data while meeting an acceptable level of accuracy.
This “Occam’s Razor” inspired approach helps restrict errors introduced by uncertainties associated with additional parameters, Lima said.
“We try to balance the number of parameters and the quality of the prediction,” Lima said. “So if a model is already good, according to your definition, why do you need to increase the complexity if it’s not giving you a better result?”
Lima started his work on the tumor models in 2011 as a Ph.D. student in his native Brazil at the National Laboratory for Scientific Computing. The research captured his long-held fascination with using mathematics to describe biology, an interest he stoked as a master’s student at São Paulo State University. There he computationally modeled the impact of temperature on the life cycle of certain insects.
During his Ph.D. studies he spent a year at ICES working with his co-advisor ICES Director Tinsley Oden, who at that time led the institute’s tumor modeling group. Thanks to support from the ICES Peter O’Donnell, Jr. Postdoctoral Fellowship in Computational Engineering and Sciences, he was able to return to the institute in 2014.
“Everyone who works with finite elements knows Dr. Oden, so that was a reason to come as a Ph.D. student,” Lima said. “The second time I came to ICES as a postdoc, it was because he really was a great person to work with and also because of all the computational resources.”
Another benefit of research at ICES is the evolution of the tumor modeling research at the institute, Lima said. Over the past couple years the group’s collaborators have grown, prompting the formation of a new ICES Center for Computational Oncology in 2016. The collaboration has expanded the team to include researchers from across The University of Texas System, including the Department of Biomedical Engineering, M.D. Anderson Cancer Center and the Dell Medical School.
“To work together in this multidisciplinary team, you must be willing to learn the other area. We are sharing biology and math, so we can check if our ideas are correct,” Lima said.
The research progress since Lima has returned is tangible, said Oden, with models evolving from sophisticated representations of cancer’s theoretical behavior to actual predictive tools.
“Ernesto developed a lot of the algorithms that are essential to this success,” Oden said. “We’ve implemented a lot of new statistical methods that allowed us to cope with uncertainties in the data and do cutting edge work in mathematics, statistics, and scientific computing. He’s been a stalwart on those subjects and is now an essential part of the team.”
The team’s research methodology has incorporated 13 tumor growth models of varying complexity, with some developed at the institute, some from other research groups. According to a paper published in Mathematical Models and Methods in Applied Sciences in October, the suite of models was able to accurately predict the growth patterns of actual tumors in rats.
Lima says that the more he learns about cancer, the more he’s amazed that the team has been able to capture the behavior of such a complex disorder.
“You see the equations, you know what every term of the equation means, but if you know how complex the tumor growth is—there’s a lot of stochastic behavior and a lot of genetic mutations—it’s amazing you can see this mathematical formulation that allows you to track this huge problem.”
From working with an array of scientists across disciplines, to seeing the tumor models emulate experiments, Lima said conducting postdoctoral research at ICES has been a formative career experience.
“I can’t think about not having been here. It’s an amazing place, it’s helped me in my development of research, and I work with specialists on tumor growth, getting real data and a chance to work with Dr. Oden who has a huge knowledge on model development,” Lima said. “It will be very important to a lot of what I’m doing from here on.”
ICES is currently accepting applications to its Post Doctoral Fellow Program. The deadline to apply is January 4, 2017.
Written by Monica Kortsha.