Partitionfinder failes with RAxML single thread version

[Karen]

Hi all,

after running PF successfully with RAxML and try out several things it again crashed (always different error messages) - the last time with this one:

INFO     | 2015-07-26 19:50:16,215 | Finished subset 201/740, 27.16 percent done
ERROR    | 2015-07-26 19:50:16,250 | The phylogeny program failed, output follows, in case it's helpful for finding the problem
ERROR    | 2015-07-26 19:50:16,251 | Warning, you specified a working directory via "-w" Keep in mind that RAxML only accepts absolute path names, not relative ones! This is RAxML version 8.2.0 released by Alexandros Stamatakis on July 9 2015. With greatly appreciated code contributions by: Andre Aberer (HITS) Simon Berger (HITS) Alexey Kozlov (HITS) Kassian Kobert (HITS) David Dao (KIT and HITS) Nick Pattengale (Sandia) Wayne Pfeiffer (SDSC) Akifumi S. Tanabe (NRIFS) Alignment has 1642 distinct alignment patterns Proportion of gaps and completely undetermined characters in this alignment: 40.72% RAxML Branch length scaler and other model parameter optimization up to an accuracy of 1.000000 log likelihood units Using 1 distinct models/data partitions with individual per partition branch length optimization All free model parameters will be estimated by RAxML GAMMA model of rate heteorgeneity, ML estimate of alpha-parameter GAMMA Model parameters will be estimated up to an accuracy of 1.0000000000 Log Likelihood units Partition: 0 Alignment Patterns: 1642 Name: No Name Provided DataType: AA Substitution Matrix: LG4X Using fixed base frequencies RAxML was called as follows: /home/kmeusema/bin/partitionfinder-develop/programs/raxml -f B -s ./analysis/phylofiles/315771d7dd5e952af5ff222161954112.phy -t ./analysis/start_tree/RAxML_result.BLTREE -m PROTGAMMALG4X -n 315771d7dd5e952af5ff222161954112_LG4X.txt -w /scratch/kmeusema/PFinder/DDA_aaDom/analysis/phylofiles -e 1.0 -O -54022.7337392497793189249932765960693359375000 -54019.5971335696303867734968662261962890625000
ERROR    | 2015-07-26 19:50:16,251 | raxml: optimizeModel.c:1634: optParamGeneric: Assertion `tr->likelihood >= initialLH' failed.

I am running a dataset wit the rcluster algortithm (only 45 taxa, ca. 300000 aa, one of the smallest one I have

Does this error refer to RAxML or PF ? I can't see a relative path, also don't know whether to post this in the PF or RAxML google group so I so both.

Any help appreciated
Karen

[Alexandros Stamatakis]

dear karen,

this looks like a raxml issue, to debug I'd need the input files and the
command line that failed,

alexis

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

--
Alexandros (Alexis) Stamatakis

Research Group Leader, Heidelberg Institute for Theoretical Studies
Full Professor, Dept. of Informatics, Karlsruhe Institute of Technology
Adjunct Professor, Dept. of Ecology and Evolutionary Biology, University
of Arizona at Tucson

www.exelixis-lab.org

[Karen]

Hi, Alexi,

Note, that I ran RAxML 8.2.0 (compiled raxmlHPC.SSE3 but renamed the binary into raxml because Partitionfinder requires this)
dataset & sh with the command line I started the analysis file & partitionfinder file I will send separately to you and Rob.

The analyses run fine up to a certain step. If Partitionfinder internally provides RAxML certain commands in each step (corrct Rob?)
Hopefully Rob Lanfer (here in cc) can be more helpful.

INFO     | 2015-07-26 19:50:16,215 | progress   | Finished subset 201/740, 27.16 percent done
ERROR    | 2015-07-26 19:50:16,250 | main       | The phylogeny program failed, output follows, in case it's helpful for finding the problem
ERROR    | 2015-07-26 19:50:16,251 | main       | Warning, you specified a working directory via "-w" Keep in mind that RAxML only accepts absolute path names, not relative ones!

what I noticed from the log file that it seems to crash using the LG4X model, the command comes from Partitionfinder:
here the command

/home/kmeusema/bin/partitionfinder-develop/programs/raxml -f B -s ./analysis/phylofiles/315771d7dd5e952af5ff222161954112.phy -t ./analysis/start_tree/RAxML_result.BLTREE -m PROTGAMMALG4X -n 315771d7dd5e952af5ff222161954112_LG4X.txt -w /scratch/kmeusema/PFinder/DDA_aaDom/analysis/phylofiles -e 1.0 -O -54022.7337392497793189249932765960693359375000 -54019.5971335696303867734968662261962890625000

ERROR    | 2015-07-26 19:50:16,251 | main       | raxml: optimizeModel.c:1634: optParamGeneric: Assertion `tr->likelihood >= initialLH' failed.

In the current run (again restarted) the Cluster admins where I am running it noticed that one RAxml run was <defunct>
(parent process Partitionfinder still runs). However this analyses is still running but not yet at the step where it crashed last time
Currently its stalled (I will also send Rob the current log file separately.

Would be great if you at the end could help us: we are currently trying unning a couple of PF analyses with large datasets (still small compared what will come)  - before we can do any tree inference

Best & thanks
Karen

[Alexandros Stamatakis]

Hi Karen,

The latest RAxML release (v 8.2.2) and the latest ExaML release (v
3.0.15), both pushed today, should fix most problems associated with LG4X.

However, this nested parameter optimization in LG4X is more complicated
and prone to get stuck in local optima than I and others originally thought.

Thus, I am only about say 70% confident that it won't get stuck in local
optima, albeit it performs better than previous versions ...

Alexis

On 27.07.2015 08:13, Karen wrote:

[Karen]

thanks, Alexis!
I will run some tests and keep you updated
k