`matUtils extract` creates invalid MAT/NWK with branch lengths of 0

[aofarrel]

In various circumstances, matUtils extract will create invalid MATs/nwks that have all branch lengths set to 0.

• For affected MATs (-o), the result is a tree that has branch lengths that do not match the number of mutations in said branches.
• For affected nwks (-t), the result is basically useless there is no list of mutations to fall back upon.

Update I've attached a script and small PB to demonstrate the issue.
proof_of_concept.txt
70.pb.txt (.txt file extension so github will upload it)

Affected commands (not exhaustive)

• matUtils extract -i -z -o (select some random number of samples to keep, output MAT)
• matUtils extract -i -s -o (select samples to keep, output MAT)
• matUtils extract -i -s -p -o (select samples to prune, output MAT)
• matUtils extract -i -s -E -o (select samples to keep, try -E as a Hail Mary even though that's for -t, output MAT)
• matUtils extract -i -s -E -t (select samples to keep, do not recalculate branch lengths, output nwk) --> This is the only situation I've found that affects -t output

Unaffected commands (also not exhaustive)

• matUtils extract -i -o
• matUtils extract -i -t
• matUtils extract -i -t
• matUtils extract -i -E -t

Specific examples

Valid input tree

Invalid MAT (per NWK inside)

Invalid NWK

Exploring an invalid MAT using BTE

NWK output in auspice and Microreact (they're trying their best, bless their hearts)

[AngieHinrichs]

@aofarrel is there anything invalid about the Newick representation aside from the missing branch lengths?

[aofarrel]

I've updated the original post with a script and PB demonstrating the issue. Previously I had issues with -t working correctly, but I can't replicate that anymore so I've removed that part.-t is still in the attached script to indicate what the nwks within the pb output should look like.

@aofarrel is there anything invalid about the Newick representation aside from the missing branch lengths?

I haven't found any direct issues, although I suspect there's going to be some issues with how matUtils condenses nodes.