Bug - pllPartitionsValidate

[Karen Meusemann]

Hi all,

refering to my question yesterday:

when I start either ML trees or BS trees with a random starttree, 

/e.g. with the comand: iqtree -s FcC_supermatrix.fas -st CODON -spp codon.best_scheme.nex -gmedian -pre codon_nt_bacoca_0.8_boot_1_20 -t RANDOM -safe -bo 20 -nt 20

It immendiatel crashes: 

e.g. with following message:

493 CODON 14 151 131 11 MGK+F1X4+R2 EOG090R0GOD

494 CODON 14 737 592 28 MGK+F3X4+R2 EOG090R03P4

495 CODON 14 441 293 20 SCHN05+F1X4+G4 EOG090R04CJ

496 CODON 14 963 668 42 MGK+F3X4+R4 EOG090R024X

497 CODON 14 211 185 8 MGK+F3X4+R2 EOG090R0AZR

498 CODON 14 255 223 16 MGK+F3X4+R2 EOG090R0DPK

Degree of missing data: 0.000

Info: multi-threading strategy over partitions

ERROR: pllPartitionsValidate

(I also tried 1.6.2 and 1.5.5 - but same error).

Does this not work with random start trees? (would be better at least for BS replicates due to recent discussions...)

Best, Karen

[Minh Bui]

Hi Karen,

This is actually a different bug which I just fixed in a beta version:

https://github.com/Cibiv/IQ-TREE/releases/tag/v.1.6.6.b

• Wrong error with codon models and -t RANDOM (reported by Karen Meusemann)

Can you please test this?

Thanks for the report.

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.

[Karen Meusemann]

Hi Minh,

ok I will test it and let you know as soon as possible!

Many thanks in advance. (I guess the LG4x/M bug will also be fixed in a version 1.6.6 later on, right?)
Karen

[Karen Meusemann]

Hi Minh,

I used now the version 1.6.6beta and did

- normal ML tree searches with parsimony start tree: they finished successfully

- ML tree searches with random start trees: they finished successfully

- BS (in batches of 20 BS replicates:

2 jobs crashed after ca. 12 hours, the others are still running (now with no error after 23 h

last lines of the ones that failed:

WARNING: Estimated model parameters are at boundary that can cause numerical instability!

BETTER TREE FOUND at iteration 1: -4181831.433

Iteration 10 / LogL: -4181833.655 / Time: 1h:16m:47s

Iteration 20 / LogL: -4181832.545 / Time: 1h:30m:46s

Finish initializing candidate tree set (1)

Current best tree score: -4181831.433 / CPU time: 3419.270

Number of iterations: 20

--------------------------------------------------------------------

|               OPTIMIZING CANDIDATE TREE SET                      |

--------------------------------------------------------------------

ERROR: iqtree.cpp:3193: virtual std::pair<int, int> IQTree::optimizeNNI(bool): Assertion `appliedNNIs.size() != 1' failed.

ERROR: STACK TRACE FOR DEBUGGING:

ERROR: 1   funcAbort()

ERROR: 2   ()

ERROR: 3   gsignal()

ERROR: 4   abort()

ERROR: 6   IQTree::optimizeNNI(bool)

ERROR: 7   IQTree::doNNISearch()

ERROR: 8   IQTree::doTreeSearch()

ERROR: 9   runTreeReconstruction(Params&, IQTree*&)

ERROR: 10   runStandardBootstrap(Params&, Alignment*, IQTree*)

ERROR: 11   runPhyloAnalysis(Params&, Checkpoint*)

ERROR: 12   main()

ERROR: 13   __libc_start_main()

ERROR: 14   _start()

ERROR:

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

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

ERROR: ***    Log file: codon_nt_bacoca_0.8_boot_r_v166b_81_100.log

ERROR: ***    Alignment files (if possible)

command that i used: 

Command: iqtree -s FcC_supermatrix.fas -st CODON -spp codon_nt_bacoca_0.8_1.best_scheme.nex -gmedian -pre codon_nt_bacoca_0.8_boot_r_v166b_81_100 -t RANDOM -safe -bo 20 -nt 20

Maybe i will see if the remaining 3 jobs finish successfully or/and crash as well and then we see how to go on?

All the best, but at least the "pllPartitionsValidate" Error is not there anymore... :)

or do you want to send me Alignment, model file and one of the failed BS logfiles?

Best, karen

[Karen Meusemann]

Hi Minh, 

all jobs succesfully finished (with the last beta version (1.6.6.c) :)

many many thanks for fixing it so quickly !!

have all a nice sunday 

Karen