error concerning dppdiv-pthreads-sse3

[Li Wang]

Dear Users

I got the newest version of pll-dppdiv https://github.com/ddarriba/pll-dppdiv uploaded on our server. When using the dppdiv-pthreads-sse3 to compute the divergence time of the tree, we got some errors. 

My command line is: ./dppdiv-pthreads-sse3 -T 28 -in dpp.data.phy -out EU.data.out -tre dpp.tree.phy -cal rel.EU.cal -n 7500000 -sf 750 -pf 1000 -npr 2 -pm 5 -bdr 25 -bda 0.3 -dphp 2 -ghp

It seems that the program ran a little while but ended up very shortly. The attached file showed the detail information of the output. 

The error showed up after couple of "rate group elements":

Rate Group Elements: (0.32225) -> 10 37 59 65 75 76 91 93 108 132 165 169 176 213 223 224 226 229 234 240 245 246 253 265 278 279 281 291 300 309 313 328 348 355 357 367 371 383 384 388 397 432 434

ERROR: Problem adding interior node to tree

Has anyone been confronted with similar error? Any suggestions or comments are very appreciated. 

Best

Li

[Tracy Heath]

Hi Li,

It seems like there might be a problem with your tree file. This error occurs when reading the Newick tree string. Are you certain that your tree is rooted and fully bifurcating? There cannot be any polytomies in the tree, and this is where this error usually comes up. Also, there needs to be a newline at the end of your tree file. 

Cheers,

Tracy

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

[Li Wang]

Hi, Tracy

Thanks for the quick reply! We are sure that the tree is rooted and bifurcating. Now I added a new line at the end of tree file and also did the dos2unix transformation of my input files. I got the program running a bit longer, but still with errors.  I tried both the dppdiv-sse3 and dppdiv-pthreads-sse3, both gave me the same error. 

Concentration parameter: 0.8349

Root height parameter: 232.2372

Speciaton parameters: d/b = 0.0000 , b-d = 25.0000

Exponential calibration parameters (DP model):  = [expected = 16.2566, gHPAlph = 2.0000, gHPBetaDP = 32.5132]

lnL = -1289672.6273

     1 -- -1289672.627 -> -1289465.718

dppdiv-sse3: PLL/evaluateGenericSpecial.c:658: evaluateIterative: Assertion `partitionLikelihood < 0.0' failed.

/opt/gridengine/default/spool/compute-32-1/job_scripts/1188396: line 12: 31536 Aborted                 (core dumped) /lustre/work/apps/pll-dppdiv-master/dppdiv-sse3 -in dpp.data.phy -out EU.data.out -tre dpp.tree.phy -cal rel.EU.cal -n 7500000 -sf 750 -pf 1000 -npr 2 -pm 5 -bdr 25 -bda 0.3 -dphp 2 -ghp

The detailed information of the output file is in the attachment. 

Best

Li

在 2013年10月11日星期五UTC-5上午10时56分34秒，Tracy Heath写道：

[Tracy Heath]

Hi Li,

That error appears to be related to the PLL. Maybe Diego, Alexis, Tomas, or Fernando can help with that one. 

Cheers!

Tracy

[Tomas Flouri]

Hi Li,

Is your dataset composed of dna data?

The failing assertion may be related to the calibration points. Can you try to run it without the calibration file and the related options and let us know of the result?

Thanks
Tomas

[王丽]

Hi, Tomas

Thanks for your reply! Yes, the dataset is composed of dna data. I tried to run without the calibration file and the related options. It didnot resolve the problem. Similar errors show up:

dppdiv-pthreads-sse3: PLL/evaluateGenericSpecial.c:658: evaluateIterative: Assertion `partitionLikelihood < 0.0' failed.

/opt/gridengine/default/spool/compute-18-28/job_scripts/1196825: line 12: 32608 Aborted                 (core dumped) /lustre/work/apps/pll-dppdiv-master/dppdiv-pthreads-sse3 -in dpp.data.phy -out EU.data.out -tre dpp.tree.phy -n 7500000 -sf 750 -pf 1000 -npr 2 -pm 5 -bdr 25 -bda 0.3

Do you have further suggestions? We appreciate your help very much!

Thanks

Li

2013/10/12 Tomas Flouri <flo...@gmail.com>

[Tomas Flouri]

Hi Li,

could you send me your tree and alignment file?

Also, what gcc version are you using?    The command is:  gcc --version

Tomas

2013/10/12 Tomas Flouri <flo...@gmail.com>

锟斤拷 2013锟斤拷10锟斤拷11锟斤拷锟斤拷锟斤拷锟斤拷UTC-5锟斤拷锟斤拷10时56锟斤拷34锟诫，Tracy Heath写锟斤拷锟斤拷

[Tomas Flouri]

Hi Li,

I tried to reproduce the error but with no luck. Do you get the following error:

dppdiv-pthreads-sse3: PLL/evaluateGenericSpecial.c:658: evaluateIterative: Assertion `partitionLikelihood < 0.0' failed.

immediately when you run DPPDiv? Or does it compute a few MCMCs (at least the first) before it crashes?

On 10/14/2013 05:50 PM, 锟斤拷锟斤拷 wrote:

2013/10/12 Tomas Flouri <flo...@gmail.com>

锟斤拷 2013锟斤拷10锟斤拷11锟斤拷锟斤拷锟斤拷锟斤拷UTC-5锟斤拷锟斤拷10时56锟斤拷34锟诫，Tracy Heath写锟斤拷锟斤拷

[王丽]

Hi Tomas

Thanks for the patience! The error was generated immediately after running DPPDiv, rather than after a few MCMCs. 

The file is in the attachment. Do you get any errors when running on your server?

Best

Li

2013/10/15 Tomas Flouri <flo...@gmail.com>

Hi Li,

I tried to reproduce the error but with no luck. Do you get the following error:

dppdiv-pthreads-sse3: PLL/evaluateGenericSpecial.c:658: evaluateIterative: Assertion `partitionLikelihood < 0.0' failed.

immediately when you run DPPDiv? Or does it compute a few MCMCs (at least the first) before it crashes?

2013/10/12 Tomas Flouri <flo...@gmail.com>

在 2013年10月11日星期五UTC-5上午10时56分34秒，Tracy Heath写道：

[Martin Stervander]

Hi Tomas and Li,

Did you ever resolve this? I have tried to run PLL-dpp-div on a dataset and run into exactly the same problem as Li did. As input I have used a bifurcating tree produced by RAxML, subsequently rooted in MEGA6 and re-exported in newick format (I also tried a version where I removed BS values, but no improvement), saved in unix and with a new line added. I have tried this both locally on my computer (Ubuntu 12.04), where my cpu allows running the SSE version, and on a CentOS server with the AVX version. (I had to modify the installation on the server according to https://groups.google.com/forum/#!msg/raxml/1SokchDLTyQ/gmnxDgYf1B4J, but I guess that whould be of no relevance.)

I called dpp-div with the command

dppdiv-pthreads-avx -in concat11426loc.fasta.reduced -out concat11426loc.red.dppdiv.out -tre RAxML_bipartitionsBranchLabels.concat11426_rax_out_rooted.nwk -cal Poecile_CIgeol.cal -T 5

where the infile is a phylip file (despite its name), the tree as described above, and the calibration file a simple one with three calibration points with uniform distributions determined by lower and upper level bounds. (Number or threads is set low while testing, and was supposed to be scaled up on the server.)

No MCMC starts, instead, invariably, the error "ERROR: Problem adding interior node to tree" is produced. I attach an error log as well as the original input tree here, and I would be very happy to provide the full dataset for you Tomas, should you have time to take a look at it.

Hoping for a solution, as I am very keen to run dpp-div on this dataset! (And with 14,500 sites, it would be nice to do it using the parallelized version...). Hopefully it's just me missing out on something fairly obvious or easy. 

Thanks a lot in advance! Best regards,

Martin

[Deren Eaton]

Hey everyone, 

I ran into this same problem recenly. I think it is a bug with parsing the newick string. When I deleted the bootstrap support values (removed node.labels from the tree using the ape package in R) PLL-dpp-div was able to run just fine. 

Deren