Sampling Local Minima

Evolutionary Search for Mapping Minima in the Protein Energy Surface


Ph.D. Defense Talk - Work by B. Olsong and A. Shehu* at ACM Conf on Bioinf and Comp Biol (BCB) 2013, AAAI-W Workshop on AI and Robotics Methods for Computational Biology 2013, Genet Evol Comp Conf (GECCO) 2013, BMC Struct Biol J 2013, IEEE BIBM-W Comp Struct Biol Workshop (CSBW) 2012, Proteome Sci J 2013, IEEE Intl Conf on Bioinf and Biomed (BIBM) 2012, Proteome Sci J 2012, and IEEE Intl Conf on Bioinf and Biomed (BIBM) 2011.

Protein energy surfaces are nonlinear and multimodal, which makes them suitable systems to study with evolutionary search/optimization algorithms. We are currently exploring such algorithms to effectively sample local minima in the protein energy surface. These minima are of relevance when studying thermodynamically-stable and semi-stable structural states that a native protein uses for its biological function or a variant employs for loss of function. Our focus is on equipping the basic algorithmic frameworks with domain-specific (biophysical) knowledge on proteins and then pursuing adaptations of the basic frameworks for an enhanced exploration capability. The large objective is to employ these algorithms to obtain a detailed characterization of the structure space and model the structure-function relationship in protein and protein-like systems.

Our work has investigated the basic Basin Hopping framework, more powerful hybrid population-based frameworks, implementation of various global and local moves in evolutionary search algorithms, and the incorporation of multi-objective optimization through Pareto-based metrics to attenuate the reliance on noisy energy functions and obtain a more diverse conformational ensemble. Details can be found in the related pages and publications.

On this Project:

  • Brian Olson

    Sameh Saleh (Undergraduate Student)

    Irina Hashmi

    Kenneth De Jong

    Amarda Shehu

This material is based upon work supported by the National Science Foundation under Grant No. 1016995 and IIS CAREER Award No. 1144106. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.