The Shehu lab is pushing the capability of evolutionary algorithms (EAs) to move beyond optimization and obtain comprehensive and detailed maps of complex, multi-modal energy landscapes.
The application focus is on multi-basin energy landscapes of proteins that switch between basins to modulate their biological activity. The EAs we design combine domain-specific insight from protein modeling and biophysics with novel algorithmic techniques to delay premature convergence and achieve a balance between exploration and exploitation with modest computational budgets.
Specifically, the image shows the ability of a novel EA not only to provide a comprehensive map of the landscape of a naturally-occurring, healthy form of a protein but also provide a map of an oncogenic variant. Side-by-side comparisons reveal the impact of the mutation on the landscape. The impact can be quantified by comparing costs of the transition between the two main basins that correspond to the known stable states of this protein. This work is among the first to allow understanding the thermodynamic and structural impact of mutations as a way of providing detail on how mutations cause dysfunction.
This work is supported by grants from the National Science Foundation (Grant Nos. 1421001 and 1144106). For more information, visit the Shehu lab