nickjcroucher / gubbins

Rapid phylogenetic analysis of large samples of recombinant bacterial whole genome sequences using Gubbins
http://nickjcroucher.github.io/gubbins/
GNU General Public License v2.0
174 stars 50 forks source link

Error: Bipartitions missing when harmonising roots between trees #360

Closed eam12 closed 11 months ago

eam12 commented 1 year ago

I've been using Gubbins for years and am a huge fan. Thank you for developing this program. Today I encountered an error I've never received before:

Bipartitions missing when harmonising roots between trees gubbins/tmplua1dxdq/coreFiltered_clean.full.1675353838.iteration_1.treefile and gubbins/tmplua1dxdq/coreFiltered_clean.full.1675353838.iteration_1.tre.rooted
The missing bipartitions are: ['00000000000000000000000000000000000000000000000000000000000000001100000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000']

I took a peek at the source code and apparently this is something that can happen when using FastTree to fit a phylogenetic model, but I'm using IQ-Tree:

run_gubbins.py --prefix gubbins --tree-builder iqtree --threads 12 --use-time-stamp coreFiltered_clean.full.aln

Any clue as to what's going on? Thanks so much!

nickjcroucher commented 1 year ago

Hmm, interesting - what version are you using?

eam12 commented 1 year ago

I've encountered the same error with both versions 3.1.6 and 3.3.

nickjcroucher commented 1 year ago

The issue can arise if there is a multifurcation in the trees - can you share the two files that are referred to in the error?


From: eam12 @.> Sent: Thursday, February 2, 2023 6:18:25 PM To: nickjcroucher/gubbins @.> Cc: Croucher, Nicholas J @.>; Comment @.> Subject: Re: [nickjcroucher/gubbins] Error: Bipartitions missing when harmonising roots between trees (Issue #360)

I've encountered the same error with both versions 3.1.6 and 3.3.

— Reply to this email directly, view it on GitHubhttps://github.com/nickjcroucher/gubbins/issues/360#issuecomment-1414170407, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AD6PWQZOF5S2CX5QGC6B633WVP27DANCNFSM6AAAAAAUPKVUZM. You are receiving this because you commented.Message ID: @.***>

eam12 commented 1 year ago

What's the best way to send them to you? They do contain some unpublished data so I would prefer to send them to you privately, if possible.

nickjcroucher commented 1 year ago

Of course - you can email me

nickjcroucher commented 1 year ago

There are two taxa on ~zero length branches that are differently positioned on the two trees - this occurs after the model fitting, which should be using IQtree. Could you try using --model-fitter raxml and check whether that fixes the issue?

noemielefrancq commented 1 year ago

Hello all,

I've ran into the same error. I'm working on a large tree with some very small branch lengths. I was using iqtree as the tree builder and the model fitter. Gubbins version 3.3.0, iqtree v2.2.3.

I can confirm changing the model fitter to raxml (--model-fitter raxml) fixed the issue!

Best, Noemie