Problem running ancestral state reconstruction

[Philip Ashton]

Hello,

I'm running ancestral state reconstruction for a large alignment (>300000 columns, > 800 taxa) and I'm encountering an error. This is the command I'm using

~/programs/iqtree-1.6.beta4-Linux/bin/iqtree -asr -s 2017.06.09.all_ours_and_desj.snp_sites.mod.fa -te 2017.06.09.all_ours_and_desj.snp_sites.fast.mod.tree -m GTR+ASC+G4 -nt 16

And the error I'm receiving is this

ERROR: phylokernelnonrev.h:1091: double PhyloTree::computeNonrevLikelihoodBranchSIMD(PhyloNeighbor *, PhyloNode *) [VectorClass = Vec4d, nstates = 4, FMA = false]: Assertion `prob_const < 1.0 && prob_const >= 0.0' failed.

The full log file is downloadable here, and the alignment (80 MB when gzipped) is here.

It predicts the state of the first node before crashing. I ran a small test dataset which completed successfully.

Any ideas?

Phil Ashton

[Bui Quang Minh]

Hi Phil,

Thanks for the report. While I will examine the problem, can you please rerun without +ASC in the model and let me know if it works? Because I suspect there is a problem with ancestral reconstruction under +ASC.

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.

--

Bui Quang Minh

Center for Integrative Bioinformatics Vienna (CIBIV)

Campus Vienna Biocenter 5, VBC5, Ebene 1

A-1030 Vienna, Austria

Phone: ++43 1 4277 74326

Email: minh.bui (AT) univie.ac.at

[Philip Ashton]

That worked! Thanks for the fix and the quick response!

[Philip Ashton]

Hi Minh,

I have a different problem with running ASR. Now I'm running it with large alignments (19 million nucleotides), much of which is invariant. The analysis crashes after reading the sequence in. I don't think it is a RAM issue, as the machine I'm working on has 128 GB.

The log file is attached.

Best,

Phil

[Bui Quang Minh]

Dear Phil,

That’s weird, can you send to my personal email the alignment for debugging purpose?

Thanks, Minh

<2017.08.10.all_l4_vs_h99.fa.log>

[Bui Quang Minh]

Dear Philip,

I could reproduce the crash using the data you sent me, and moreover got a backtrace:

ERROR: STACK TRACE FOR DEBUGGING:

ERROR: 1   funcAbort()

ERROR: 2   ()

ERROR: 3   pllAlignmentRemoveDups()

ERROR: 4   IQTree::initializePLL(Params&)

ERROR: 5   runTreeReconstruction(Params&, std::string&, IQTree&, ModelCheckpoint&)

ERROR: 6   runPhyloAnalysis(Params&, Checkpoint*)

ERROR: 7   main()

So it means that the bug is in the (old) phylogenetic likelihood library (PLL) that IQ-TREE uses for parsimony computation. Some users already reported this crash when the alignment is very long.

To avoid this, please add option -t PARS into the command line (to switch to IQ-TREE parsimony kernel). Let me know how it works.

Thank you, Minh

<2017.08.10.all_l4_vs_h99.fa.log>

[Philip Ashton]

Sorry it took me so long to respond. Yes, this fix works well. Thank you!