BUG? mixed binary/nucleotide analysis

[Paul Frandsen]

Dear Minh,

I am working on a combined binary/nucleotide tree search. I am running version 1.6.7 on 64 bit Linux. I followed the instructions on the IQtree website for setting up the nexus partition scheme file and split the alignment into two phylip files as instructed. When I start the search (iqtree-1.6.7-Linux/bin/iqtree -s aa_nuc.phy -pre aa_nuc -spp aa_part.nex -m MFP -nt 12 -safe -bb 1000 -bnni -g Trichop.edit10.TLcons.tree), it reads the alignments fine, but then fails with the following error:

Creating fast initial parsimony tree by random order stepwise addition...

308.958 seconds, parsimony score: 28591 (based on 1900 sites)

NOTE: ModelFinder requires 105 MB RAM!

Testing GTR2+G,GTR+F+G on supermatrix...

ERROR: phylotree.cpp:760: size_t PhyloTree::getBufferPartialLhSize(): Assertion `num_threads > 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: aa_nuc.log

ERROR: ***    Alignment files (if possible)

Is there anything clear that I might have done wrong? I'm happy to send log files and alignments along to your personal email.

Many thanks,

Paul

[Minh Bui]

Hi Paul, That looks weird. Can you send me the input data for debugging?

Thanks,

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.

[Cristian Beza]

Hi Migh,

I had the same problem when I ran it with Linux. When I ran the analysis in widows version 1.5.5 it would start the analysis but it would crash after an n number of Bootstrap replicates. and will bive the following error:  Tree Does not start with oppening bracket

[Juan Carlos Zamora Señoret]

I have also the same problem. 

It seems that you need to specify the appropriate model for each partition, and specifically a binary model for the binary dataset (e.g., MK), otherwise it cracks. 

e.g.

#nexus

begin sets;

    charset part1 = DNA1.fas: *;

    charset part2 = DNA2.fas: *;

    charset part3 = Binary.fas: *;

charpartition mine = GTR+F+I+G4:part1, GTR+F+I+G4:part2, MK+ASC:part3;

end;

This will work. 

But the real problem is that the command "-TESTMERGE" will not work. So, I do not see how it is possible to have, at the same time, the binary dataset with the binary model (and perhaps the ASC correction if you need it) and test the partitioning scheme with PartitionFinder for the remaining datasets...

Also, I do not know how to combine partitions included as independent alignments (or parts of alignments) in the NEXUS file, since between partitions you can only use "," to separate partitions or ";" to end the block. I mean, you cannot use "part1+part2", or "part1_part2", etc.

The only solution would be to run a previous analysis without the binary dataset to test the partitioning scheme for the other datasets, and then manually arrange the partitions to comply with that partitioning scheme. Very inconvenient, particularly if you have partitions by codon position, since you would have to split each codon position, then mergin them with others, etc. Or merge everything into a single long alignment, and then call the partitions as usually done for other software, but then the great versality of IQ-Tree for using independent alignments would be lost.

Did you reach any other conclusion on this issue?

Best wishes,

Juan Carlos

[Minh Bui]

Hi Juan,

ModelFinder should work with mixed data. Did you use the latest version 1.6.8? There was a crash with mixed data type with ModelFinder, which was fixed in 1.6.8. If the problem still persists, please send me the log file.

Thanks,

Minh

[Juan Carlos Zamora Señoret]

Dear Minh,

Well, now I am normally (see at the end!) getting the following message (command line: "-spp partitions.nex -m TESTMERGE"):

ERROR: Numerical underflow (lh-branch). Run again with the safe likelihood kernel via `-safe` option 

An with the -safe option ("-spp partitions.nex -m TESTMERGE -safe") I am getting:

ERROR: \Users\minh\Dropbox\iqtree-git\model\partitionmodelplen.cpp:142: virtual double PartitionModelPlen::optimizeParameters(int, bool, double, double): Assertion `cur_lh > tree_lh - 1.0 && "partition rate opt reduces LnL"' failed.

This application has requested the Runtime to terminate it in an unusual way.

Please contact the application's support team for more information.

ERROR:

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

The DNA dataset alone always work well, the binary dataset alone always work well, and the combined dataset when specifying a model for each partition and not testing the partitioned scheme always work well. I am getting the error only when I want to test the partitioning scheme in the combined dataset via "-m TESTMERGE", like in the previous version but with a different error.

HOWEVER, after several trials with "-spp partitions.nex -m TESTMERGE", it randomly works. I would say around one of each 4-6 trials works. Any idea why? When it works, the tree is perfectly fine, everything as expected.

Best wishes,

Juan Carlos

[Minh Bui]

Hi Juan,

Can you please send me via personal email the data set for debugging purpose? Also the log file would be good to quickly get what’s going on.

Thank you,

Minh

[Cristian Beza]

Dear Minh,

I had the same bug. Would you be so Kind to help me with this. Thank you so much

Minh

To unsubscribe from this group and stop receiving emails from it, send an email to iqt...@googlegroups.com.

[Minh Bui]

Hi Cristina, 

Did you try the latest version (1.6.10)? I believe this problem was fixed…

Minh