Static Bayesian Modeling of Biological Time-Series Data

Recent research into reconstructing biological networks has examined the use of dynamic Bayesian networks to model time-series data. While intuitively appealing, dynamic Bayesian network modeling makes assumptions about the properties of time-series data which may not hold for sparsely sampled datasets. This work argues that static Bayesian networks may be a more appropriate model for such datasets. Several exploratory results supporting this argument are presented on datasets previously cited in the dynamic Bayesian learning literature.