Improvements resulting from testing the model on Ebola:
Introduced negative binomial observation model which suited noisy ebola data better
Made some performance improvements with a better poisson sampler
Tree segmenting now happens during data load in, not in every MCMC step
Max epidemic size no longer supported due to fixed poisson sampler (this is a good thing)
No longer throws an error if the output folder already exists; just makes a duplicate with the system date and time
Still more to do to improve phylo acceptance rate on the ebola data but I don't think this is in the model code, I think maybe in the model parameterisation
Improvements resulting from testing the model on Ebola:
Still more to do to improve phylo acceptance rate on the ebola data but I don't think this is in the model code, I think maybe in the model parameterisation