Bayesian model inference and parameter estimation for biological models

http://sorgerlab.github.com/bayessb

BayesSB is an algorithm and software suite for estimating parameter distributions in ODE-based models of cellular biochemistry and for discriminating between models having different numbers of unknown parameters. The algorithm is described in detail in Eydgahi et al. Properties of cell death models calibrated and compared using Bayesian approaches. Mol Syst Biol (in review).

The procedure returns joint probability distributions for model parameters and makes it possible to compute uncertainty for model-based predictions based on error in the data and the non-identifiability of model parameters. Bayesian methods also make it possible to compute the odds ratio for competing models having different numbers of parameters.

BayesSB depends only on numpy for its core functionality. However matplotlib is also needed if you wish to use the provided plotting routines in bayessb.plot. PySB is our preferred tool for building models which integrate well with BayesSB, but it is not required.

The code has extensive inline documentation. Once you have installed the package, run pydoc bayessb and pydoc bayessb.plot to view it.

For a fairly simple example of BayesSB usage, see the script examples/earm/fit_1_3_standalone.py.

• We should absolutely use the PyMC package instead of rolling our own MCMC code.
• Better support of non-PySB models, e.g. SBML import.
• Implementation of the Bayes factor calculations from the MSB paper.