search

ManchesterBioinference / burstInfer

Using a dynamic memory-adjusted Hidden Markov Model to infer global and single-cell transcriptional parameters
Apache License 2.0
5 stars 1 forks source link

How to run the full HMM model? #2

Open hongpengzhou opened 2 years ago

hongpengzhou commented 2 years ago

The released code is the proposed cpHMM in this paper. But can I use this code to run the HMM by just setting PERMITTED_MEMORY = 2^W, where W means the window size.

jbqbb commented 2 years ago

The exact model is included as part of the package, but there's a separate function to use for the full model (if you try and implement it by changing the PERMITTED_MEMORY value to the 'actual' value then the code will crash, as the model will try to use 2 * the full state space).

To run the exact model use the following code snippet:

learned_parameters = demoHMM.exact_EM(initialised_parameters, n_steps, n_traces, eps, seed_setter)

After commenting out the line of code to run the truncated model. The exact_EM function is part of the HMM class - you want to use this instead.

I should probably include a full example in the examples folder.