Nodes that are not samples and not added by the local tree building have pretty arbitrary node times. It would be helpful in the short term to have a method to roughly retime these nodes (until we implement it in the tree building process and rerun building).
Something crude like scaling the node's time relative to the number of mutations it carries and the number of mutations on the path from its parent to the node's closest descendant sample (or something) would probably give a much better approximation than we have now. If we do this forwards in time it could be done in a single pass.
Nodes that are not samples and not added by the local tree building have pretty arbitrary node times. It would be helpful in the short term to have a method to roughly retime these nodes (until we implement it in the tree building process and rerun building).
Something crude like scaling the node's time relative to the number of mutations it carries and the number of mutations on the path from its parent to the node's closest descendant sample (or something) would probably give a much better approximation than we have now. If we do this forwards in time it could be done in a single pass.