TITLE:
Development and Application of a Modified Genetic Algorithm for Estimating Parameters in GMA Models
AUTHORS:
José A. Hormiga, Carlos González-Alcón, Néstor V. Torres
KEYWORDS:
Parameter Estimation, Genetic Algorithms, GMA Models, Model Calibration, Inversion Methods, JAK2/STAT5 Pathway Model
JOURNAL NAME:
Applied Mathematics,
Vol.5 No.16,
August
29,
2014
ABSTRACT:
In this work we introduce a modified version of the
simple genetic algorithm (MGA) and will show the results of its application to
two GMA power law models (a general theoretical branched pathway system and a
mathematical model of the amplification and responsiveness of the JAK2/STAT5
pathway representing an actual, experimentally studied system). The two case
studies serve to illustrate the utility and potentialities of the MGA method
for concerning parameter estimation in complex models of biological
significance. The analysis of the results obtained from the application of the
MGA algorithm allows an evaluation of the potentialities and shortcomings of
the proposed algorithm when compared with other parameter estimation algorithm such
as the simple genetic algorithm (SGA) and the simulated annealing (SA). MGA
shows better performance in both studied cases than SGA and SA, either in the
presence or absence of noise. It is suggested that these advantages are due to
the fact that the objective function definition in the MGA could include the
experimental error as a weight factor, thus minimizing the distance between the
data and the predicted value. Actually, MGA is slightly slower that the SGA and
the SA, but this limitation is compensated by its greater efficiency in finding
objective values closer to the global optimum. Finally, MGA can lead to an
early local optimum, but this shortcoming may be prevented by providing a great
population diversity through the insertion of different selection processes.