BUG: IQ-TREE CRASHES WITH SIGNAL SEGMENTATION FAULT (core dumped)

[Simon Hellemans]

Dear developers,

I have first been running IQ-TREE v1.6.12 on my alignment without specifying any partitions and the run went fine.

However, when running now by specifying partitions, I get segmentation fault error.

Please find beginning and end of the log file below.

Thanks in advance for your reply.

Best,

Simon

------------------------------------------------------- BEGINNING -------------------------------------------------------

IQ-TREE multicore version 1.6.12 for Linux 64-bit built Aug 15 2019

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

Heiko Schmidt, Dominik Schrempf, Michael Woodhams.

Host:    deigo010915.oist.jp (AVX2, FMA3, 503 GB RAM)

Command: iqtree -s supermatrix.nex -spp all_partitions.nex -nt AUTO -ntmax 24 -mem 10G

Seed:    566713 (Using SPRNG - Scalable Parallel Random Number Generator)

Time:    Tue Feb 23 21:22:29 2021

Kernel:  AVX+FMA - auto-detect threads (128 CPU cores detected)

Reading partition model file all_partitions.nex ...

Reading alignment file supermatrix.nex ... Nexus format detected

------------------------------------------------------- END -------------------------------------------------------

Loading 5 partitions...

Subset  Type    Seqs    Sites   Infor   Invar   Model   Name

1               320     2933    1664    916             rRNA

2               326     1922    975     716             tRNA

3               326     3939    2019    1561            coding_first

4               326     3939    2127    1493            coding_second

5               326     3938    2499    1097            coding_third

Degree of missing data: 0.003

Info: multi-threading strategy over partitions

Create initial parsimony tree by phylogenetic likelihood library (PLL)... 0.582 seconds

NOTE: ModelFinder requires 1197 MB RAM!

NUMBER OF THREADS FOR PARTITION FINDING: 5

Testing GTR+F+G on supermatrix...

GTR+F+G4 / LnL: -1053664.626 / df: 658 / AIC: 2108645.251 / AICc: 2108699.413 / BIC: 2113725.950

Selecting individual models for 5 charsets using BIC...

 No. Model        Score       Charset

   1 GTR+F+R5     137405.745  tRNA

   2 GTR+F+I+G4   278265.320  rRNA

   3 GTR+F+R5     485433.650  coding_first

   4 GTR+F+R5     514751.069  coding_second

   5 GTR+F+R5     655716.423  coding_third

Full partition model BIC score: 2078001.114 (LnL: -1035466.819  df:727)

Partition information was printed to all_partitions.nex.best_scheme.nex

Partition information in Raxml format was printed to all_partitions.nex.best_scheme

All model information printed to all_partitions.nex.model.gz

CPU time for ModelFinder: 11042.569 seconds (3h:4m:2s)

Wall-clock time for ModelFinder: 3254.475 seconds (0h:54m:14s)

NOTE: 574 MB RAM (0 GB) is required!

Measuring multi-threading efficiency up to 24 CPU cores

ERROR: STACK TRACE FOR DEBUGGING:

ERROR: 

ERROR: *** IQ-TREE CRASHES WITH SIGNAL SEGMENTATION FAULT

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

ERROR: ***    Log file: all_partitions.nex.log

ERROR: ***    Alignment files (if possible)

/var/spool/slurmd/job5848028/slurm_script : ligne 57 : 95727 Erreur de segmentation  (core dumped)iqtree -s $ALIGNMENT -spp ${WORKDIR}/${ANALYSIS_ALL}.nex -nt AUTO -ntmax ${SLURM_CPUS_PER_TASK} -mem 10G

[Simon Hellemans]

Hi,

I think I narrowed it to down to multi-threading issues. I restarted the run without specifying -nt and -ntmax with the below command and the run could complete:

iqtree -s $ALIGNMENT -spp ${WORKDIR}/${ANALYSIS_ALL}.nex -redo

Best,

Simon

[Minh Bui]

Hi Simon,

Does it work, if you specify something like -nt 10, instead of -nt AUTO? That means, tell IQ-TREE the specific number of threads, instead of asking IQ-TREE to determine it.

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 view this discussion on the web visit https://groups.google.com/d/msgid/iqtree/b6d2d7ad-058d-47a0-9b43-c007d98b8bb9n%40googlegroups.com.

[Simon Hellemans]

Dear Minh,

Thank you for your reply.

Specifying the number of threads was actually my first test. I specified -nt to 24, which gave me a lot of the same warning followed by the same crash reported above:

WARNING: Number of threads seems too high for short alignments. Use -nt AUTO to determine best number of threads.

I am currently running it again with -nt set to 10 and will let you know how that goes.

Best,

Simon

[Simon Hellemans]

Hi Minh,

The job with -nt 10 crashed as reported with -nt AUTO.

Best,

Simon

[Minh Bui]

Hi Simon,

I look more carefully now. You have only 5 partitions. And IQ-TREE even said:

NUMBER OF THREADS FOR PARTITION FINDING: 5

So can you rerun with -nt 5?

Anyway please send me the data, I’ll try to fix this bug. IQ-TREE shouldn’t have crashed, but do something about small datasets like this to limit the number of threads.

Minh

To view this discussion on the web visit https://groups.google.com/d/msgid/iqtree/56783c6b-ac85-4b39-81d5-7a55b11ed6d1n%40googlegroups.com.

[Simon Hellemans]

Hi Minh,

I just tried with -nt 5 but the same crash occurs unfortunately.

I just sent you the data for you to have a look at it.

Best,

Simon

[Simon Hellemans]

Dear Minh,

Did you get any time for looking into this issue?

Best,

Simon

[Francesco Cicconardi]

Hi,

I'm having the same problem.

F

[Thomas Wong]

Hi, Simon (and Francesco),

I have tried IQ-TREE version v.2.2.2.5 (available at https://github.com/iqtree/iqtree2/releases) on your data set with your command and an option -nt 5. It could work and finish with no crash. This version has an issue fixed when the input data set with partitions of multiple data types.

Thomas

[Eleanor Stroud]

Dear developers,

I've been having the same issue as well. The code I'm using is " iqtree-2.2.6-Linux/bin/iqtree2 -s seqtab6.fasta.mafft.trimal.phy -nt 1 -m GTR+F+ASC+R10 -bb 10000"

We've managed to optimise the analysis to one thread as suggested by -nt AUTO, and the model is the one selected by the model picker. I've updated to the most recent version (2.2.2.6) but it's still giving me this error message;

Alignment was printed to seqtab6.fasta.mafft.trimal.phy.uniqueseq.phy

For your convenience alignment with unique sequences printed to seqtab6.fasta.mafft.trimal.phy.uniqueseq.phy

CHECKPOINT: Initial tree restored
Generating 10000 samples for ultrafast bootstrap (seed: 384176)...

NOTE: 6850 MB RAM (6 GB) is required!
CHECKPOINT: Model parameters restored, LogL: -1377479.785
Wrote distance file to...
Computing ML distances based on estimated model parameters...

ERROR: STACK TRACE FOR DEBUGGING:
ERROR:
ERROR: *** IQ-TREE CRASHES WITH SIGNAL SEGMENTATION FAULT
ERROR: *** For bug report please send to developers:

ERROR: ***    Log file: seqtab6.fasta.mafft.trimal.phy.log

ERROR: ***    Alignment files (if possible)

/tmp/slurmd/job1369199/slurm_script: line 4: 3828994 Segmentation fault      (core dumped) iqtree-2.2.6-Linux/bin/iqtree2 -s seqtab6.fasta.mafft.trimal.phy -nt 1 -m GTR+F+ASC+R10 -bb 10000

Please could you advise how to resolve this?

Eleanor

[Thomas Wong]

Dear Eleanor,

Thanks for reporting the issue. Can you send me the log and phy data files so we can check?

Thanks,

Thomas

To view this discussion on the web visit https://groups.google.com/d/msgid/iqtree/f8c19720-9eb8-4ad7-8a15-79c6fad27b98n%40googlegroups.com.