beast-dev / tracer

Posterior summarisation in Bayesian phylogenetics
https://beast.community/tracer
GNU Lesser General Public License v2.1
55 stars 17 forks source link

Tracer 1.7.1 does ont open outfiles of G-Phocs #153

Closed gscabanne closed 3 years ago

gscabanne commented 4 years ago

Hi, we have realized that ver 1.7 does not open outfiles of G-Phocs, a software to estimate population genetic parameters with genomics data-sets Gphocs web is>http://compgen.cshl.edu/GPhoCS/

Previous versions of Tracer (i.e., 1.5 and 1.6) do open the outfiles. This has been consisten across different outfiles and platforms.

What happenes is a) the outfiles have two blank spaces instead of one space separating the last two columns of data. Changing two to one space helped to al least open the file, but Tracer was not able to properly read the parameters, and presented very weird tracer graphs. I was ont able to figure out what is the cause of the problem. So, I am using now Tracer 1.6 or 1.5 and there not any problem to open these outfiles. Apparently, there has been a change from ver 1.6 to 1.7 that precluded Tracer to open G-Phocs outfiles Did anyone else experience this issue?

Best

MarineBio5 commented 4 years ago

Hi,

I experienced this issue too. For those (like me) who struggled to find an older version it is available here: https://github.com/beast-dev/tracer/releases/tag/v1.6 Best

walterxie commented 3 years ago

Any example files to reproduce this issue?

gscabanne commented 3 years ago

Hi, you may try with the attached one. I currently am using version 1.6. This issue is now well known among gphocs users.best GUSTAVO SEBASTIÁN CABANNE Investigador-Researcher Museo Arg. de Cs. Naturales "Bernardino Rivadavia", Buenos Aires, Argentina Tel: 54 11 4982 6595 int 186 Associate Editor of Ornithology Research -Springer- (former Revista Brasileira de Ornitología), and Molecular Phylogenetics and Evolution -Elsevier-

http://gscabanne.wixsite.com/science http://www.conicet.gov.ar/new_scp/detalle.php?keywords=&id=32339&datos_academicos=yes

On Thursday, April 29, 2021, 08:38:34 PM GMT-3, Walter ***@***.***> wrote:  

Any example files to reproduce this issue?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

walterxie commented 3 years ago

@gscabanne there is no attachment from your previous posts.

walterxie commented 3 years ago

It will be appreciated if anyone could provide an example file for this issue.

gscabanne commented 3 years ago

NOw there is. GUSTAVO SEBASTIÁN CABANNE Investigador-Researcher Museo Arg. de Cs. Naturales "Bernardino Rivadavia", Buenos Aires, Argentina Tel: 54 11 4982 6595 int 186 Associate Editor of Ornithology Research -Springer- (former Revista Brasileira de Ornitología), and Molecular Phylogenetics and Evolution -Elsevier-

http://gscabanne.wixsite.com/science http://www.conicet.gov.ar/new_scp/detalle.php?keywords=&id=32339&datos_academicos=yes

On Friday, April 30, 2021, 01:19:16 AM GMT-3, Walter ***@***.***> wrote:  

@gscabanne there is no attachment from your previous posts.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

FatihSarigol commented 3 years ago

mcmc-trace_1.log Hi, old post, but I dont see a file in the last post so I attach mine.. So I am using 1.6 version fine, but wanted to try 1.7 because Demographic Reconstruction analysis doesn't respond on 1.6 for this file again from Gphocs, so that can be another issue to report here. Yeah so 1.7 doesnt open this file. thanks

walterxie commented 3 years ago

Hi @FatihSarigol, thank you very much to provide the example log file. There are 2 problems in this log file:

  1. The empty column was added at the last 3rd column. The screen shot is attached. After I read Gphocs manual https://raw.githubusercontent.com/gphocs-dev/G-PhoCS/master/GPhoCS_Manual.pdf, I think it is a bug of Gphocs. Tracer parser will ignore the empty column, as long as the header (parameter names) is in a correct format, such as the log file you provided. But since Tracer 1.7, we added a validation method after loading, which was using a different code to read, so this empty column made Tracer crashed. I have updated that code to use the same parsing process, which will handle this matter.

  2. An extra white space is added in the beginning of all values. I did not check all columns, but at least theta_NORWAY column. I am not sure if it was done for any other purposes, but it looks unnecessary and problematic. So I add a trim method to remove any leading and trailing white space from the value string.

Screenshot from 2021-08-17 14-35-10

walterxie commented 3 years ago

You log file works in the latest code. The bug-fix will be available in the next release. Alternatively, you can build the jar using the latest code in the repository. I will leave this issue open for several days.

FatihSarigol commented 3 years ago

Hey Walter, Thank you for checking and figuring out the tiny issue! Before going for building the new jar, I wanted to try to fix the problems you pointed out in Gphocs' output file.

This code below did the job, which fixes both of the problems, it removes those spaces in front of the values and fixes the field shift issue in the header as well (I write it here in case anyone else would find this post before the new release) :

awk -v OFS="\t" '$1=$1' mcmc-trace_1.log > fixed.log

and so it now opens with version 1.7.2

Can I ask another question here, Walter; should I be able to run Demographic Reconstruction or any of the other analyses with this file or am I missing something? Such as do I need a specific trace log for Population Size or does it work from any of the samples I already have in the file? Because for me none of the analyses work really.

Thanks a lot!

walterxie commented 3 years ago

Hi Fatih, we cannot control the behavior and output of other software. There is already a warning message in the GUI panel:

Screenshot from 2021-08-19 10-32-17

Personally, I suggest you to use BEAST or LPhy.