RAxML as tree estimator causes SATe to crash

[Shabnam Zaman]

Hi everyone,

I apologise if this question has been addressed before, but for some reason every time I try to run SATe using RAxML as the tree estimator, the program immediately crashes. As such, I can only use FastTree - but this is not ideal for my purpose. I've checked the list of published bugs and evidently this is a known issue when using Windows 32-bit systems... however, I'm running a 64-bit system instead. Any advice on this matter and how to proceed would be highly appreciated. Thanks in advance!

[Tandy Warnow]

Hi Shabnam,

I'll let Mark Holder and his group comment on this, but here are suggestions for how to deal with this.

• You might just run FastTree within SATe, and then *after* you get the alignment from SATe you can run RAxML on it. In other words, use RAxML only after you get the alignment from SATe.
• Alternatively, you might try PASTA. This is the most recent method within the SATe family of alignment/tree estimation methods, and can run on larger datasets (up to 1,000,000 sequences so far). It's possible that PASTA will not have a problem with what you are trying to do. If you know how to run SATe, you'll find running PASTA to be extremely easy. See https://github.com/smirarab/pasta for the github repository, and Mirarab, S., Nguyen, N. Guo, S., Wang, L., Kim, J. and Warnow, T.. PASTA: Ultra-Large Multiple Sequence Alignment for Nucleotide and Amino-Acid Sequences. Journal of Computational Biology (2014)

Best wishes,

Tandy Warnow

On Wed, Mar 8, 2017 at 6:52 AM, Shabnam Zaman <shabnam...@gmail.com> wrote:

Hi everyone,

I apologise if this question has been addressed before, but for some reason every time I try to run SATe using RAxML as the tree estimator, the program immediately crashes. As such, I can only use FastTree - but this is not ideal for my purpose. I've checked the list of published bugs and evidently this is a known issue when using Windows 32-bit systems... however, I'm running a 64-bit system instead. Any advice on this matter and how to proceed would be highly appreciated. Thanks in advance!

--
--
You received this message because you are subscribed to the Google Groups "SATe User" group.
To post to this group, send email to sate...@googlegroups.com
To unsubscribe from this group, send email to sate-user+unsubscribe@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sate-user?hl=en

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

[Shabnam Zaman]

Hi Tandy,

Thanks a lot for the prompt response. Ideally, I'd like to construct a tree within a single program run; as such, I want to avoid the two-step / post-processing aspects if at all possible. Moreover, my dataset consists of only a few hundred rather short sequences, so I'd be surprised if that's the root of the problem. Thank you for the PASTA suggestion though - I'll look into that soon. :)

On Wednesday, March 8, 2017 at 2:05:14 PM UTC+1, tandy wrote:

Hi Shabnam,

I'll let Mark Holder and his group comment on this, but here are suggestions for how to deal with this.

• You might just run FastTree within SATe, and then *after* you get the alignment from SATe you can run RAxML on it. In other words, use RAxML only after you get the alignment from SATe.
• Alternatively, you might try PASTA. This is the most recent method within the SATe family of alignment/tree estimation methods, and can run on larger datasets (up to 1,000,000 sequences so far). It's possible that PASTA will not have a problem with what you are trying to do. If you know how to run SATe, you'll find running PASTA to be extremely easy. See https://github.com/smirarab/pasta for the github repository, and Mirarab, S., Nguyen, N. Guo, S., Wang, L., Kim, J. and Warnow, T.. PASTA: Ultra-Large Multiple Sequence Alignment for Nucleotide and Amino-Acid Sequences. Journal of Computational Biology (2014)

Best wishes,

Tandy Warnow

On Wed, Mar 8, 2017 at 6:52 AM, Shabnam Zaman <shabnam...@gmail.com> wrote:

Hi everyone,

I apologise if this question has been addressed before, but for some reason every time I try to run SATe using RAxML as the tree estimator, the program immediately crashes. As such, I can only use FastTree - but this is not ideal for my purpose. I've checked the list of published bugs and evidently this is a known issue when using Windows 32-bit systems... however, I'm running a 64-bit system instead. Any advice on this matter and how to proceed would be highly appreciated. Thanks in advance!

--
--
You received this message because you are subscribed to the Google Groups "SATe User" group.
To post to this group, send email to sate...@googlegroups.com

To unsubscribe from this group, send email to sate-user+...@googlegroups.com

For more options, visit this group at http://groups.google.com/group/sate-user?hl=en

---
You received this message because you are subscribed to the Google Groups "SATe User" group.

To unsubscribe from this group and stop receiving emails from it, send an email to sate-user+...@googlegroups.com.

[Tandy Warnow]

SATe might have an option for a *final RAxML analysis* -- so you might be able to achieve what I suggested (using FastTree within the loop, and then RAxML at the end) by setting up the SATe run appropriately.  I am pretty sure that PASTA enables that.

Best wishes,

Tandy

To unsubscribe from this group, send email to sate-user+unsubscribe@googlegroups.com

For more options, visit this group at http://groups.google.com/group/sate-user?hl=en

---
You received this message because you are subscribed to the Google Groups "SATe User" group.

To unsubscribe from this group and stop receiving emails from it, send an email to sate-user+unsubscribe@googlegroups.com.

[Shabnam Zaman]

Okay, I think that's what I'm going to go with: FastTree as a tree estimator, followed by an extra RAxML search in a post-processing step. Thanks a lot!

[Shabnam Zaman]

Hello again,

Once again, I've been trying to run SATe using RAxML as the tree estimator directly, rather than during post-processing. However, this is the error I keep getting:

SATe INFO: Creating a starting tree for the SATe algorithm...
SATe INFO: Performing initial alignment of the entire data matrix...
SATe INFO: Performing initial tree search to get starting tree...
Worker dying.  Error in job.get_results = Traceback (most recent call last):
  File "sate\scheduler.pyo", line 53, in worker
  File "sate\scheduler.pyo", line 256, in get_results
  File "sate\scheduler.pyo", line 201, in wait
Exception: SATe failed because one of the programs it tried to run failed.
The invocation that failed was: 
    "D:\software\SATe\bin\raxmlp.exe" "-m" "GTRGAMMAI" "-n" "default" "-q" "C:\Users\xxx\.sate\mrna292\tempvcf0sx\init_tree\tempraxml5_mg7q\partition.txt" "-s" "C:\Users\xxx\.sate\mrna292\tempvcf0sx\init_tree\tempraxml5_mg7q\input.phy" "-T" "4"

SATe ERROR: SATe is exiting because of an error:
SATe failed because one of the programs it tried to run failed.
The invocation that failed was: 
    "D:\software\SATe\bin\raxmlp.exe" "-m" "GTRGAMMAI" "-n" "default" "-q" "C:\Users\xxx\.sate\mrna292\tempvcf0sx\init_tree\tempraxml5_mg7q\partition.txt" "-s" "C:\Users\xxx\.sate\mrna292\tempvcf0sx\init_tree\tempraxml5_mg7q\input.phy" "-T" "4"



Could you kindly advise me on how to overcome this error?

Thanks in advance,

~ Shabnam.

[Shabnam Zaman]

UPDATE: Using FastTree as the tree estimator followed by an extra RAxML search in a post-processing step also fails. The "raxmlp.exe" executable refuses to work under any circumstances. Any suggestions on how to proceed would be greatly appreciated.

[Mark Holder]

Hi,

One possibility is that virus-prevention software is blocking raxml from running. I have seen cases in which a user needs to launch the software tool (in this case D:\software\SATe\bin\raxmlp.exe ) from the prompt (e.g. the PowerShell.exe or CMD.exe) once. If you do that and get a pop-up window that warns you about running software downloaded from the internet, then you can click a choice indicating that you are willing to let raxmlp run. Then the next time sate runs, raxml should not fail at this step.

That may not be the cause of this crash, but I have seen that behavior before. Unfortunately, I do not know of any standard interface to cause the anti-virus popup window to show up when you are running SATe.

all the best,

Mark

To unsubscribe from this group, send email to sate-user+unsubscribe@googlegroups.com

For more options, visit this group at http://groups.google.com/group/sate-user?hl=en

---
You received this message because you are subscribed to the Google Groups "SATe User" group.

To unsubscribe from this group and stop receiving emails from it, send an email to sate-user+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Mark Holder

mtho...@gmail.com

[Shabnam Zaman]

Hi Mark,

Thank you for the suggestions! Unfortunately, the problem remains unresolved... I whitelisted raxmlp.exe in my antivirus program, and even tried to run it in Windows 7 compatibility mode (I run Windows 10 64-bit). Each time, the program crashed as soon as it reached the post-processing step. Launching raxmlp.exe from the command prompt returned the following message (i.e., no pop-up):

The number of threads is currently set to 0
Specify the number of threads to run via -T numberOfThreads
NumberOfThreads must be set to an integer value greater than 1

Kind regards,

~ Shabnam.

Mark Holder

mtho...@gmail.com

[Tandy Warnow]

What happens if you run SATe without RAxML at all on your data. Does it still trigger a problem? If so, then it's not RAxML that's the problem but something else.

Sorry if you already answered this...

Tandy

To unsubscribe from this group, send email to sate-user+unsubscribe@googlegroups.com

For more options, visit this group at http://groups.google.com/group/sate-user?hl=en

---
You received this message because you are subscribed to the Google Groups "SATe User" group.

To unsubscribe from this group and stop receiving emails from it, send an email to sate-user+unsubscribe@googlegroups.com.

[Mark Holder]

I agree with Tandy that it would be nice to know if raxml runs on your machine. You may need to download an example data set from the raxml site.

The -T argument is required, and should be followed by small integer greater than 0. the number of cores on your machine is probably the best choice for efficiency, but for a test of raxml:

  -T 1 

should work. The last error that you reported is resulting from 0 threads being an illegal value.

I think you'll also need a path to an input file and a -m argument to specify the model.

The fact that you are getting a message from raxml makes me think that this is not an anti-virus issue (as I originally expected), but seeing the error message (or knowing that it succeeded outside of SATe would help).

all the best,

Mark

--

Mark Holder

mtho...@gmail.com

[Pejvak Moghimi]

Hi all,

I have been having exactly the same issue; whether I run the RaxML as the main tree estimator or as the post-processing.

Fasttree works fine.

Cheers,
Pejvak.

[Pejvak Moghimi]

Hi Mark,

RaxML, independently of sate, works on both of my machines, but sate RaxML options both fail on either of the machines.

Best,
Pejvak.

[Shabnam Zaman]

Apologies for the delayed response. I tried running raxmlp.exe from the command line again (this time specifying a -T value), which prompted the pop-up you referred to. I allowed the program to run... but despite this, I received the exact same error message at the relevant step when trying to run SATe again afterwards. Also, in case I didn't mention it previously: FastTree on its own as the tree estimator works perfectly fine. It's only when I try to use RAxML (as either the tree estimator, or in conjunction with FastTree as a post-processing search) that SATe crashes, due to the aforementioned raxmlp.exe file.

At present I'm trying to carry out the same analysis in PASTA via VirtualBox (i.e., using RAxML as the tree estimator). It hasn't crashed yet, so I'll post an update when I have more news in this regard. As for running RAxML independently of SATe: I normally use a web-server and so haven't tested out RAxML locally on my computer, but as Pejvak has pointed out, the issue does not seem to be with RAxML itself.

Kind regards,

~ Shabnam.

On Thursday, March 9, 2017 at 1:47:52 PM UTC+1, Mark Holder wrote:

Mark Holder

mtho...@gmail.com

[Shabnam Zaman]

Okay, I can now confirm that RAxML works fine in PASTA, as both the tree estimator as well as for the post-processing search. So it looks like the problem is with SATe (at least, on Windows OS).

Kind regards,

~ Shabnam.

[Mark Holder]

Thanks for following up. That really helps. I'll take a look at the version of RAxML bundled with PASTA (and the newest version of RAxML). It is possible that the version of RAxML that we distribute with SATe relies on some Windows DLL or Windows OS feature that is no longer supported.

I'll need to find a Windows machine with a recent OS to test on...

Thanks again, and sorry for the difficulties you experienced.

all the best,

Mark

To unsubscribe from this group, send email to sate-user+unsubscribe@googlegroups.com

For more options, visit this group at http://groups.google.com/group/sate-user?hl=en

---
You received this message because you are subscribed to the Google Groups "SATe User" group.

To unsubscribe from this group and stop receiving emails from it, send an email to sate-user+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Mark Holder

mtho...@gmail.com

[Shabnam Zaman]

Please note that I'm running PASTA on Ubuntu via VirtualBox, whereas all the SATe runs were on my host machine (Windows 10 OS). I've never tried running SATe on a Linux system, so maybe Windows is the culprit here after all.

Thanks again for all your suggestions!

Kind regards,

~ Shabnam.