search

OpenTreeOfLife / treemachine

Source tree graph database
Other
16 stars 6 forks source link

question about edges to licas during TAG creation. #161

Closed mtholder closed 9 years ago

mtholder commented 9 years ago

no need to respond to this if it makes sense to you. I can look through the code. Just thought I'd post it because I found it surprising. There is a new test (and I refactored the test harness some to put the output of tests in the testtag dir). That is pushed to https://github.com/OpenTreeOfLife/treemachine/tree/test-refactor

The new test is ranksforextras and its tree1 and tree3 are identical to the tree1 and tree2 for the nontrivialaugmenting test. I just added a new tree in between them that disagrees about the placement of the "extra" taxa E and F. The test passes (it prefers tree2 to tree3), which is great. But I was surprised that in nontrivialaugmenting trees 1 and 2 share a lot of nodes: nontriv

but in ranksforextras tree3 is using nodes that are out on their own, and not traversed by tree1 (note tree to color mapping is not consistent): ranksforextras

updated tree2 was mislabeled

mtholder commented 9 years ago

the second figure from is https://github.com/OpenTreeOfLife/treemachine/tree/test-refactor/test-synth/ranksforextras

josephwb commented 9 years ago

tree1: (((A,B),C),D); tree2: ((A,B),((E,F),C)); tree3: ((((A,E),B),F),C);

For the second figure (not the first, i think). I believe that in the first figure, tree1 = tree1, and tree2 = tree3.

kcranston commented 9 years ago

When I draw this out, I agree with Mark about n9 == n17, i.e. mrca(tree3,(A,E,B)) and mrca(tree1,(A,B)) should be the same node.

blackrim commented 9 years ago

This should be alleviated with deleting the trees and re-adding them. We know that tree order matters in input and deleting and re inputting is the way around that. This is default for many of the analyses (discussed by Cody and Jonathan the other day). The commands for that are java blahblah pgdelind graphdblocation sourcename then you add again and should be mapped to both of the lica nodes. Joseph can perhaps answer more on IRC

On Wed, Feb 4, 2015 at 1:55 PM, Karen Cranston notifications@github.com wrote:

When I draw this out, I agree with Mark about n9 == n17, i.e. mrca(tree3,(A,E,B)) and mrca(tree1,(A,B)) should be the same node.

— Reply to this email directly or view it on GitHub https://github.com/OpenTreeOfLife/treemachine/issues/161#issuecomment-72915102 .

mtholder commented 9 years ago

calling pgdelind and addnewick to remap seems to help:

reprocess

thanks for the help.