Anthony Wilder Wohns
Anthony Wilder Wohns
A script to systematise qc checks of tree sequence quality would be really helpful. This could be a script called via the CLI 'tsinfer-qc' or something like that. We've found...
A medium to long-term goal is to see if one could piece apart the precise signal of error in inferred tree sequences. Sites with hundreds of mutations are almost certainly...
The root node of the tgp tree sequence has hundreds of thousands of child edges, a large proportion of which (>50%) are sample edges. This can cause issues with tsdate....
@hyanwong, Gil and I have recently discussed details of a "windowed" GNN, which would use either span- or time-based windows to calculate the nearest neighbours of a set of focal...
We currently can't date an unsimplified tree sequence with dangling nodes. We might want to think about how to do this, the problem is that there isn't a clear prior...
The accuracy of tree sequences inferred with `tsinfer` can be improved by using `tsdate` estimated dates. This is very straightforward and should be added to the docs: ``` inferred_ts =...
@hyanwong has worked out a method for exhaustively enumerating all possible combinations of node times. This allows calculation of the exact likelihood of a particular set of node times given...
Dating accuracy is severely decreased with inferring tree sequences with unary nodes. We're probably not handling something correctly. Unittests with simple examples of unary nodes may help.
Logging
Should add logging to the cli to keep user updated on what's happening. More debug statements would be good too
We should treat epsilon as the probability of two nodes being in the same timeslice. We need to work out the probability of that occurring