For the moment only static Bayesian Networks are supported but we are currently working on the development of Dynamic Bayesian Networks

These are the available features :

Distributions

• Multinomial Distributions (discrete)
• Gaussian Distributions (multivariate)

Inference Engines

• Junction Tree Inference (only for discrete networks)
• MCMC Sampling Engine (also for hybrid networks discrete/continuous)

Learning

• ML (Maximum Likelihood) Learning is supported for any kind of distribution (complete data)
  
• EM (Expectation Maximization) Learning is supported only for discrete distributions (should work soon for gaussians and hybrid networks)
• Structure Learning is supported only for discrete distributions (complete or incomplete data) and should work soon with gaussians and hybrid BNs
  

Other Features

• Import/Export bayesian networks using the XBN format http://research.microsoft.com/dtas/bnformat/
• Graphical representation of the network and distributions

Future features :

• Dynamic Bayesian Networks and inference for DBNs
  
• More performant inference and learning engines
• Mixture of Gaussians
  
• ...

For more information on future version planning, see the dedicated page : Future Version planning

If you need some feature for your work, we strongly encourage you to implement it within OpenBayes (that is, of course, that you are planning to use Python as well). It is much easier to implement something within an already existing platform that has been tested by many users, than to implement it yourself from scratch. We guarantee to provide help and assistance to new developers in order to maintain a coherence inside the code and help people work faster.

As an example, I will cite the work done by a member for implementing the S-EM algorithm (structure Learning with incomplete data) : Although this is quite a complicated algorithm, it took less than two months to completely implement it.