ESTIMATIONOF NEURONAL SIGNALINGMODEL PARAMETERS USING DETERMINISTICAND STOCHASTICINSILICOTRAININGDATA:EVALUATIONOF FOUR PARAMETER ESTIMATIONMETHODS

Thisstudy evaluates parameter estimation methodology inthecontext ofneuronal signaling networks. Basedon theresults ofaprevious study, four parameter estimation methods, Evolutionary Programming, Genetic Algorithm, Multistart, andLevenberg-Marquardt, areselected. Allthe reaction rate constants ofthetest case, theprotein kinase C (PKC)pathway model, areestimated using theselectedfour methods. Theestimations aredonewithbotherroranddisturbance free training datafromdeterministic insilico simulations andwithmorerealistic training data fromstochastic insilico simulations. Theresults showthat inoverall theevolution basedalgorithms perform well. However, there isaclear needforfurther development, especially whenutilizing morerealistic training data.