LPT-pre-2010-22 BibTeX
@MISC{LPT-pre-2010-22,
AUTHOR = {T. Quaiser and Anna Dittrich and Fred Schaper and M. M\"{o}nnigmann},
TITLE = {{Model simplification work flow applied to a model of JAK-STAT signaling }},
LPTKey = {LPT-pre-2010-22},
}
Tom Quaiser, Anna Dittrich, Fred Schaper, Martin Mönnigmann:
Model simplification work flow applied to a model of JAK-STAT signaling
The 11th International Conference on Systems Biology October 2010, Edinburgh, Scotland, 10-15.10.2010
Abstract:
One major issue in systems biology is the identification of dynamic models for biological pathways. When constructing such a model, it is tempting to incorporate all known interactions between species of the pathway, which results in models with a large number of unknown parameters. Fortunately, unknown parameters need not necessarily be measured directly, but some parameter values can be estimated indirectly by fitting the model to experimental data. However, parameter fitting, or, more precisely, maximum likelihood parameter estimation, can only provide valid results, if the complexity of the model and the amount and quality of data available are in balance with one another. If this is the case the model is said to be identifiable for the given data. If a model turns out to be unidentifiable, two steps can be taken. Either additional experiments need to be conducted in order to increase the information content of the data, or the model has to be simplified.
In this contribution we focus on a systematic procedure for model simplification, which proceeds along the following steps: 1) estimate the parameters of the actual model for the data given, 2) create an identifiability ranking for the estimated parameters [1], and 3) simplify the model based on the results from 2).
These steps need to be applied iteratively, until the resulting model is identifiable. For step 1 we use multi-start parameter estimation to mitigate the problem of convergence to local optimality. We use a sampling based parameter variance analysis to backup identifiability results used in step 3 and in the stopping criterion of the procedure. In contrast to related work in systems biology, we do not suggest simplifying a model by fixing parameters to arbitrary values, but change the structure of the model based on biologically reasonable assumptions.
We apply the proposed approach to a model of early signaling events in the JAK-STAT pathway. With a real systems biology project in mind, we pay special attention to envision only experiments that can actually be performed in a biochemical laboratory using currently available tools. The resulting model is identifiable and shows the best tradeoff between goodness of fit and model complexity. Furthermore, in agreement with the identifiability results, the estimated parameters of the resulting model have low variances for multiple estimation runs starting in a neighborhood of the optimal estimate.
Keywords:
systems biology, identifiability, parameter estimation, model reduction, model identification, model comparison