search

tskit-dev / tsdate

Infer the age of ancestral nodes in a tree sequence.
MIT License
19 stars 10 forks source link

Mutation metadata mean and variance for unphased singleton mutations #414

Open hyanwong opened 3 months ago

hyanwong commented 3 months ago

An unphased singleton mutation could be on either the maternal or the paternal edge. Does the mean and variance saved in the mutation metadata reflect this possibility (i.e. is X % of the weight taken from the gamma on one edge and (100-X) % from the gamma on the other edge? Or is the mean in the mutation metadata conditioned on having correctly phased the singleton?

nspope commented 3 months ago

The mean in the metadata integrates over possible phase, so is the weighted version you describe.

hyanwong commented 3 months ago

Perfect, thanks for clarifying Nate. I'll add that to the documentation for release when the singleton phasing stuff is in.