Usage
  • 14 views
  • 3 downloads

Optimization of Ion channel recordings through analysis of multiple fittings

  • Author / Creator
    Luchko, Aaron C
  • We develop an approach for optimizing Hidden Markov model representations of voltage-gated ion channels that addresses the issues of topology determination and poorly performing optimization algorithms. Developing accurate models of neurological processes is a major goal of computational neuroscience, but creating accurate models of voltage-gated ion channels is a difficult task. Noisy data, a large range of potential topologies, and large numbers of parameters make machine optimization very difficult and topology comparison techniques unreliable. We attempt to address the unreliability of the optimization process through multiple fittings. We then analyze the sets of fitted models with a new metric designed to measure consistency in the behaviour of the hidden states. When combined with the LogLikelihood this indicates whether the model has the complexity necessary to fit the data. We then design a protocol based around the creation of multiple fitted models that utilizes this metric both as a guide for further fittings and a way to identify a selection of suitable models and topologies. We apply the metric to five sets of simulated data and two pairs of live recordings of voltage-gated K+ channels. On the simulated data the described protocol generated a range of topologies that successfully captured the correct topology in all but one of the simulated trials where it underestimated the topology required. Applied to the live data the procedure performed well on one channel type, for the other results were impacted by the difficulty of the optimization problem. In general the procedure and metric performed well but were limited by the ability of the optimizer to deliver a range of high quality solutions.

  • Subjects / Keywords
  • Graduation date
    2014-11
  • Type of Item
    Thesis
  • Degree
    Master of Science
  • DOI
    https://doi.org/10.7939/R33N20N9Z
  • License
    This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.
  • Language
    English
  • Institution
    University of Alberta
  • Degree level
    Master's
  • Department
    • Department of Computing Science
  • Supervisor / co-supervisor and their department(s)
    • Wong, Ken (Computing Science)
    • Jones, Kelvin (Physical Education and Recreation)
  • Examining committee members and their departments
    • Gallin, Warren (Biology)