Changelog
---------

4.0.0
New interface: Hidden Markov models can be now fitted to backgrounds.

3.1.2
Bugfix: When Poisson distribution is used for sampling the number of motif instances and the number of instances = 0, the output is the same as the background sequence. Used to throw an error.

3.1.1
New feature: sequence letter probabilities are also provided in a npz format as an output along with the sampled final_sequence.fasta