nmut approximation via rounded expectation can be inaccurate in edge cases.

[croth1]

CCMgen/ccmpred/sampling/__init__.py

Line 37 in d874668

nmut = int(clade.branch_length * mutation_rate * ncol)

The current implementation uses the rounded expectation over the full sequence as the number of mutations nmut . This can be problematic in edge cases (sequences very short or mutation rate very low).

A better alternative in these cases would be drawing nmut probabilistically:

nmut = np.random.poisson(lam=clade.branch_length*mutation_rate, size=ncol).sum()