BUG: concordance factors

[Aaron Liston]

Wonderful to see the addition of concordance factors!  I gave it a spin with 52 taxa and 14,265 gene trees. The gene trees were previously computed with raxml-ng, and species tree with Astral3. I was very happy to obtain the gCF & sCF results as  node labels, but the program failed when preparing the stats table. Also, why does the program report 53 taxa, not 52?  Here is the log:

IQ-TREE multicore version 1.7-beta6 for Linux 64-bit built Nov 19 2018

Developed by Bui Quang Minh, Nguyen Lam Tung, Olga Chernomor,

Heiko Schmidt, Dominik Schrempf, Michael Woodhams.

Host:    groot.cgrb.oregonstate.local (AVX, FMA3, 503 GB RAM)

Command: iqtree -t Junmin2018.CDS-14265.raxml.boot_10.Astral3.tre --gcf Junmin2018.CDS-14265.raxml-ng.best.tre -p alignments-only --scf 100 --prefix Junmin2018.iqtree.concord -nt 30

Seed:    410851 (Using SPRNG - Scalable Parallel Random Number Generator)

Time:    Wed Dec  5 20:23:31 2018

Kernel:  AVX+FMA - 30 threads (64 CPU cores detected)

...

Reading tree Junmin2018.CDS-14265.raxml.boot_10.Astral3.tre ...

53 taxa and 103 branches

Reading tree(s) file Junmin2018.CDS-14265.raxml-ng.best.tre ...

14265 un-rooted tree(s) loaded

Trees have different number of taxa

Computing gene concordance factor...

5.069 sec

Computing site concordance factor...

1137.705 sec

Tree with concordance factors written to Junmin2018.iqtree.concord.cf.tree

Tree with branch IDs written to Junmin2018.iqtree.concord.cf.branch

ERROR : Could not find neighbors of node 43

ERROR: node.cpp:176: Neighbor *Node::findNeighbor(Node *): Assertion `0' failed.

ERROR: STACK TRACE FOR DEBUGGING:

ERROR:

ERROR: *** IQ-TREE CRASHES WITH SIGNAL ABORTED

ERROR: *** For bug report please send to developers:

ERROR: ***    Log file: Junmin2018.iqtree.concord.log

ERROR: ***    Alignment files (if possible)

Any ideas?  Thank you,  Aaron

[Minh Bui]

Hi Aaron, 

Thanks for giving it a try and finding the problem this quick. I haven’t tested the combination of ASTRAL tree + raxml gene trees. Thus, can you please send me via personal email the two input trees files for debugging purpose? 

Regarding

but the program failed when preparing the stats table. Also, why does the program report 53 taxa, not 52?  Here is the log:

I think ASTRAL printed a rooted tree, and the root is treated as one extra taxon in IQ-TREE. I think that could have caused the crash. Nevertheless, I’ll find out once seeing your data.

Cheers

Minh

--
You received this message because you are subscribed to the Google Groups "IQ-TREE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to iqtree+un...@googlegroups.com.
To post to this group, send email to iqt...@googlegroups.com.
Visit this group at https://groups.google.com/group/iqtree.
For more options, visit https://groups.google.com/d/optout.

[nicola...@gmail.com]

Hello,

I have tried too the new gCF analyses. Same as Aaron i used an Astral 3 species tree and gene trees generated by iqtree (1466). Same my species tree as 122 leaves but the analyses report 123. But the analysis was performed 9only gCF). But I was confused about some results as I have node labels with a local posterior probability (Astral lpp) of 1 and a gCF of 0; which is also in contradiction with PhyParts results (e.g. a node can have a lPP of 1; phypart 1448 gene concordant and  17 not concordante; gCF of 0).
This could be due to error with the leaves number?

Here is the output report:

IQ-TREE multicore version 1.7-beta6 for Linux 64-bit built Nov 19 2018
Developed by Bui Quang Minh, Nguyen Lam Tung, Olga Chernomor,
Heiko Schmidt, Dominik Schrempf, Michael Woodhams.

Host:    frontend8 (AVX2, FMA3, 125 GB RAM)
Command: ./iqtree -t Ptero_nuc1466_Astral.tre --gcf ./Genes_tree/loci.treefile --prefix concord -nt 4
Seed:    68569 (Using SPRNG - Scalable Parallel Random Number Generator)
Time:    Wed Dec 12 16:41:12 2018
Kernel:  AVX+FMA - 4 threads (48 CPU cores detected)

Reading tree Ptero_nuc1466_Astral.tre ...
123 taxa and 243 branches
Reading tree(s) file ./Genes_tree/loci.treefile ...
1466 un-rooted tree(s) loaded

Trees have different number of taxa
Computing gene concordance factor...

1.041 sec
Tree with concordance factors written to concord.cf.tree
Tree with branch IDs written to concord.cf.branch
Concordance factors per branch printed to concord.cf.stat
Date and Time: Wed Dec 12 16:41:14 2018

is there a way to fix the problem?

Thanks

Nicolas

[Minh Bui]

Hi Nicolas,

There was no crash, so right now I can’t think that’s a problem. Anyway I have posted version 1.7-beta7 that fixed issue with rooted trees:

https://github.com/Cibiv/IQ-TREE/releases

Let me know how that works.

FYI: I have seen dataset, where bootstrap support for a branch is 100% but that branch only occurred once among the gene trees. So case like this is not totally unexpected. See also a very nice blog post by Rob Lanfear: http://www.robertlanfear.com/blog/files/concordance_factors.html

Cheers

Minh

[nicola...@gmail.com]

Hi Minh,

Thank you very much for your message.

I have downloaded the last version.

The new version work correctly and report the correct number of leaves. I have tested using as input a Astral species tree or the concatenated ML tree.

Yes for the backbone nodes of my tree (which looks like as an hard polytomy) I have local Posterior Probability (statistic calculated by Astral 3) very high but gCF null. When I will have bootstrap values of my concatenated ML tree I will look if i got very high BS/very low gCF.

Thanks for the link to the blog very useful.

Best,

Nicolas