Structure and Structure Threader [Estimated Ln Prob of Data]

[Vaibhav Mahilang]

Dear Developers and Everyone,

I am using structure (Windows Command Line Version) for my analysis.

Currently, I have Windows 11, 8 GB RAM. Which I used for a few trial runs with the expected population (K) based on real-life assumption, 1000 burn-in, and 1000 repetition. Everything worked very well. Results are also aligned with the expected outcome.

Now, for actual analysis, I have access to a server PC with Windows 10, 128 GB RAM, and 16 threads, which is available solely for this specific analysis. As expected I am using structure threader for using the capabilities of server pc.

Since I have switched to Structure Threader I have started receiving a few errors, which I haven't gotten to fix by myself. The parameters to run the analysis and errors are as follows :

Parameters - Structure Threader
 
Mode - run
Klist - 17 18 19 (Real life population number - 18)
Repetitions each K - 10
Threads - 16
Log - 1

Structure mainparams

MAXPOPS - 18
NUMIND - 424
NUMLOCI - 68436
BURNIN + NUMREPS - 100 (Why Low? It is a trial run)
POPDATA - 1
POPFLAG - 1

Structure extraparams
NOADMIX - 0
USEPOPINFO - 1
INFERALPHA - 1
RANDOMIZE - 0 (Structure Threder sets it to 0 by itself )

ERRORS

1 Structure harvester cannot read lnPr(X|K) Estimated Ln Prob of Data

2 Evanno method cannot be applied 

3 Cannot generate plots 

To figure out the issue I tried to run the structure threader with different parameters and also tried to check the results. Here are my observations :

-  Results do not contain Estimated Ln Prob of Data (Explains Error 1 and 2)

-  Tried (No test for best K)  still cannot generate plots 

                    - Error  - value error for plotting (See attached picture - NoKTest.png)

- Tried ( No plots) 

                    - Error 1 and 2 (as expected) (See attached picture - NoPlotTest.png)

- Tried (No test for best K + No plots) 

                    - Logfile,f, and q file produced for every repeat

                    - Plots have to be generated manually

My question is did I miss something that caused the ln probability to not generate?

What should I do to fix this?

I apologize for the long thread, I wanted to give a clear idea of my analysis.

Any help is appreciated.

[Francisco Pina Martins]

Dear Vaibhav,

First of all, thank you very much for your detailed exposition of the problem you are running into! It's really through.

I'm the developer of Structure_threader. Let's try to get you up and running. =-)

From what you show, I suspect that this is a combination of parameters that will cause STRUCTURE to produce output files slightly different from what Structure_threader expects to find. If this is so, I'd like to get it fixed as soon as possible. In order to speed testing, can I please request that you post the exact command you used to run Structure_threader, as well as your `mainparams` and `extraparams` files, please? I will attempt to replicate your run locally and figure out exactly where the issue lies.

Thank you!

Francisco

[Vaibhav Mahilang]

Dear  Francisco,

I appreciate your reply.

I am sending the commands and attaching the mainparams and extraparams files here.

########################################################################################################################
ppath = r'x:\xxxxx\xxxxxxxxxxx\xxxxxxx\xxxxxxx\xxxxxxxx\xxxxxxxxx\ALLSCRIPT\STRUCTURE'
os.chdir(f'{ppath}')
########################################################################################################################
Ks = '17 18 19'
Rs = '10'
Ts = '16'
infileST = r' x:\xxxxx\xxxxxxxxxxx\xxxxxxx\xxxxxxx\xxxxxxxx\xxxxxxxxx\ALLSCRIPT\STRUCTURE\PB_MB_M.recode_new_filtered.strct_in'
outfileST = r' x:\xxxxx\xxxxxxxxxxx\xxxxxxx\xxxxxxx\xxxxxxxx\xxxxxxxxx\ALLSCRIPT\File\Resultstruct'
Spath = r' x:\xxxxx\xxxxxxxxxxx\xxxxxxx\xxxxxxx\xxxxxxxx\xxxxxxxxx\ALLSCRIPT\STRUCTURE\structure.exe'
Mpath = ppath + r'\mainparams'
Epath = ppath + r'\extraparams'
PECMD = 'params'
PEPATH = r'-o' + ppath
########################################################################################################################
# RUN MODE ONLY

THREADER = r' x:\xxxxx\xxxxxxxxxxx\xxxxxxx\xxxxxxx\xxxxxxxx\xxxxxxxxx\Scripts\structure_threader.exe'
TCMD = 'run'
KN = '-Klist ' + Ks
RN = '-R ' + Rs
TN = '-t ' + Ts
IP = '-i ' + infileST
OP = '-o ' + outfileST
SP = '-st ' + Spath
TEST = '--no_tests 1'
# PLOT = '--no_plots 1'
# EO = '--extra_opts '
# M = '-m ' + Mpath
# E = '-e ' + Epath
# PP = '--params ' + ppath
LOG = '--log 1'
os.system(f'{THREADER} {TCMD} {SP} {IP} {OP} {KN} {RN} {TN} {TEST} {LOG}')

######################################################################################################################## 

I hope that is enough.

Please do let me know if you need anything else.

I would not be able to share the actual file (It must not be needed I believe), Because of confidentiality.

Merry Christmas and Happy New Year.

Vaibhav Mahilang

[Francisco Pina Martins]

Dear Vaibhav,

In fact you seem to have a combination of options in your parameter files I have not tested. But I have been unable to generate a dataset that will work with them.

So, may I kindly request that you send me an anonimized file with your data, please? The first 10 individuals (just name them Samp01 to Samp10) and 100 loci should suffice for debugging. You can send this data privately to my email if you don't feel comfortable sharing it in public.

Thank you.

Best,

Francisco

[Francisco Pina Martins]

Dear Vaibhav,

With the files you have kindly provided me, I was able to reproduce the issues you encountered. Errors 1 and 2 were caused by insufficient "numreps". Increasing the numreps value to at least 10000 in the `mainparams` file solved both those errors.

Error 3, however was still occurring. This was caused by combination of the parameter "usepopinfo" being set to 1, but having some individuals with an attributed population (0). Thanks to your input, I was able to find it and correct it.

I have released Structure_threader version 1.3.11 (it's already in pypi) which should have the fix builtin. Please let me know if it solved the problem for your whole dataset as it did for the mockup you sent me.

Best,

Francisco

[Vaibhav Mahilang]

Dear Francisco,

Thanks a lot for the solution and for fixing the issues.

We have had some issues with the server so I couldn't test it, but for now, the server has been fixed and now I can use it again.

I have started the threader program now with more than 10000 NUMREPS, and as it is working for you in the mockup file, it should also work well in my whole dataset. I will let you know as soon as I have any output.

In the meantime, there are a few questions (Might be silly) I want to know about :

1 What is the significance of replicates (-R) in structure threader, and does it have any dependencies with NUMREPS? How many replicates are absolutely necessary? Will this replicate's value Let's say 10000 be the same as NUMREPS 10000? 

Can I skip (or put -R 1), when I have a fixed K number for my every analysis? or should I do --no_tests 1?

2 See the picture attached SNAP.png, when using replicates -R = 1 and K's = 16,17 and 18, the structure threader only runs in a single thread for each 16,17,18 K's for 10000 NUMREPS. so in total only using 3 threads. What do I do so that the threader can use all the threads available?

3 I understand that each line shown in the Python console (in picture SNAP.png),  is running 10000 NUMREPS in the background (Meaning that the structure threader is using structure in the background ). Is it in any way possible to see each NUMREPS in the Python console, as we were able to see via the structure program? ( attached picture structure.png)

Thanks a lot

Vaibhav

[Francisco Pina Martins]

Dear Vaibhav,

Let me try to address your questions in order:

1 What is the significance of replicates (-R) in structure threader, and does it have any dependencies with NUMREPS? How many replicates are absolutely necessary? Will this replicate's value Let's say 10000 be the same as NUMREPS 10000? 

Can I skip (or put -R 1), when I have a fixed K number for my every analysis? or should I do --no_tests 1?

The variable NUMREPS in the mainparams file, refers to the number of MCMC iterations in each STRUCTURE run. The number of required iterations varies with the dataset, it’s hard to provide a straightforward answer. But I recommend taking a close look at section 3.3 of STRUCTURE’s manual. The -R option in Structure_threader refers to how many runs you want to perform for each value of K (you need multiple runs to estimate the “best” K using Evanno’s method).

2 See the picture attached SNAP.png, when using replicates -R = 1 and K's = 16,17 and 18, the structure threader only runs in a single thread for each 16,17,18 K's for 10000 NUMREPS. so in total only using 3 threads. What do I do so that the threader can use all the threads available?

Structure_threader is a wrapper program for STRUCTURE. As such it cannot make single runs work with multiple cores (only STRUCTURE would be able to do that). What you can do to leverage your multicore system is run more values of K and/or add replicates with -R.

3 I understand that each line shown in the Python console (in picture SNAP.png),  is running 10000 NUMREPS in the background (Meaning that the structure threader is using structure in the background ). Is it in any way possible to see each NUMREPS in the Python console, as we were able to see via the structure program? ( attached picture structure.png)

Although you can’t look at the NUMREPS information “live” as in STRUCTURE, you can run Structure_threader with --log 1 option which will create a log file with that information for each replicate that gets run.

Hope this helps,

Best,

Francisco

--
You received this message because you are subscribed to a topic in the Google Groups "structure-software" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/structure-software/E_zevPCM-a8/unsubscribe.
To unsubscribe from this group and all its topics, send an email to structure-softw...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/structure-software/981cc395-32e0-429e-9a94-99eb0f833825n%40googlegroups.com.

​

[Vaibhav Mahilang]

Dear Francisco,

Thank You for your detailed answers, they have been very helpful.

Along with these current answers, I appreciate fixing the previous issues, as in the new analysis I have been able to run the program smoothly now. The combination of parameters are working very well.

With this, I am successfully able to continue my analysis, except some errors I received. I am attaching those errors here. Can you have a look at them?

ERROR 3 - There's 2 structure runs that are exited with an error and are produced in 2 log files, where I could check the error. (picture : error 3 1)

I tried to look for the 2 log files to look for the error, I found that structure_threader did not produce these files. (picture : error 3 2)

Also, can you explain more about, How only STRUCTURE can be able to make a single run use multi-core? I tried to use only STRUCTURE and it is using only a single core.

Thanks a lot

Kind Regards

Vaibhav Mahilang

[Francisco Pina Martins]

Dear Vaibhav,

Thank You for your detailed answers, they have been very helpful.

Along with these current answers, I appreciate fixing the previous issues, as in the new analysis I have been able to run the program smoothly now. The combination of parameters are working very well.

Happy to know you have made some progress.

With this, I am successfully able to continue my analysis, except some errors I received. I am attaching those errors here. Can you have a look at them?

I had never seen Structure_threader fail only on specific replicates of a repeated analysis…

From the screenshots it seems you were able to perform all runs simultaneously on your machine. K18 replicates are likely to have been the last to finish. Since they errored out and no log was written I have only one suspicion. Perhaps you have run out of storage space on the machine? If you are 100% sure you did not, can you please share the command you used to run Structrure_threader?

Also, can you explain more about, How only STRUCTURE can be able to make a single run use multi-core? I tried to use only STRUCTURE and it is using only a single core.

I am not aware of any way to do this. All the wrappers I know (strauto, parallelstructure) do the same as Structure_threader - make each separate and independent launch in a different OS thread. Allowing STRUCTURE to paralelize independent runs would probably require re-writing STRUCTURE to support it.

Best,

Francisco

To view this discussion on the web visit https://groups.google.com/d/msgid/structure-software/759f6127-b81d-42ff-9646-260b9dc02595n%40googlegroups.com.

​

[Vikram Chhatre]

Correct. STRUCTURE will have to be rewritten to parallelize it. The best we can do is to use one of the wrappers, as Francisco mentioned.

You received this message because you are subscribed to the Google Groups "structure-software" group.
To unsubscribe from this group and stop receiving emails from it, send an email to structure-softw...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/structure-software/048f08e7-d581-4e01-a98b-b8ee1f62e1dc%40gmail.com.

[Vaibhav Mahilang]

Dear Francisco and Vikram,

Thank you both for the answer.

For Francisco: 

1) I am 100% sure that I did not run out of storage. I have 128 GB RAM with 1 TB of storage, and in my experience, until now structure_threader has not used any more than 10-12% of RAM ever in my analysis yet. I am attaching the command I have used in the file ST_CMD.

2) With all the fixes, I went ahead and started my full analysis, with 6874 individuals from 137 populations, and 68436 loci, where 24 animals have POPDATA 0 and POPFLAG 0. I received the error of "Error in assigning memory (not enough space?)" (See picture Error 4 4)(also ATTCHAED LOG FILE, including COMMANDS in  K137_rep1.stlog). I tried to check the memory usage, at the time of the structure_threader run, and I saw that it is only taking ~2 GB of space.

3) When running structure_threader, the Python console displays the file locations and other parameters. In the console, I found a parameter -D, which I have not provided. Can you explain what is it? (See picture Error 4 5)

For Vikram: This is regarding the issue 2) (mentioned above for Francisco) :

Just to make sure that I understand the issue, I ran the structure standalone, and I received the same error, so I, Believe that the issue is based on structure only. (see picture Error 4 2). I also checked that it is not the issue of the Python host I am using, even the Windows command prompt output is the same. (see picture Error 4 3)

Is it something you come across often?, how can I go past this?

I know about fastStructure, for large datasets, but I am not sure if that is the solution to this issue.

Kind Regards

Vaibhav

[Francisco Pina Martins]

Hi again Vaibhav,

1) I am 100% sure that I did not run out of storage. I have 128 GB RAM with 1 TB of storage, and in my experience, until now structure_threader has not used any more than 10-12% of RAM ever in my analysis yet. I am attaching the command I have used in the file ST_CMD.

STRUCTURE won’t typically stress RAM memory usage, so what you are seeing is all within expectations. What I was suggesting was running out of storage space, not RAM memory. But let’s take a deeper look.

2) With all the fixes, I went ahead and started my full analysis, with 6874 individuals from 137 populations, and 68436 loci, where 24 animals have POPDATA 0 and POPFLAG 0. I received the error of "Error in assigning memory (not enough space?)" (See picture Error 4 4)(also ATTCHAED LOG FILE, including COMMANDS in  K137_rep1.stlog). I tried to check the memory usage, at the time of the structure_threader run, and I saw that it is only taking ~2 GB of space.

Despite STRUCTURE complaining of memory problems, browsing the list suggests that when that message comes up it is related to unusual input file issues. Can you try to replace POPDATA 0 and POPFLAG 0 with a < 0 integer, and see if that solves it?

3) When running structure_threader, the Python console displays the file locations and other parameters. In the console, I found a parameter -D, which I have not provided. Can you explain what is it? (See picture Error 4 5)

The -D parameter is the seed. If you don’t provide one, Structure_threader will automatically create one for you to make sure your work is reproducible.

Just to make sure that I understand the issue, I ran the structure standalone, and I received the same error, so I, Believe that the issue is based on structure only. (see picture Error 4 2). I also checked that it is not the issue of the Python host I am using, even the Windows command prompt output is the same. (see picture Error 4 3)

Yes, it looks that way. Hopefully setting a POPFLAG/POPDATA value < 0 solves it. Otherwise I'm kind of at a loss.

I know about fastStructure, for large datasets, but I am not sure if that is the solution to this issue.

It is a workaround, at best, since fastStructure does not have the same kind of features as STRUCTURE (such as providing POPDATA/POPFLAG information).

Best,

Francisco

To view this discussion on the web visit https://groups.google.com/d/msgid/structure-software/a8fcf726-d99b-4427-97dd-dade67fb11bcn%40googlegroups.com.

​

[Vaibhav Mahilang]

Dear Francisco,

Thank You.

I realized the issue, I noticed that in my system, the structure runs are still running with the 32-bit version.

I expected that after installing structure_threader via pip/pip3, a 64-bit version of the structure would be compiled/stored/installed? but it is not. as mentioned by you in the answers to one of the questions before in this group. or maybe I am not able to find it? maybe it does not download 64-bit in Windows?

Also, I am not able to download the 64-bit structure directly via the website, the link doesn't work. But Again you have given the updated link in the group already for the source code where I could compile it.

Along with the source code of the 64-bit structure,  you have also provided a helper script of structure_threader, which can download and compile the structure. Does this script also work on Windows? I tried to run the script, but it does not seem to do anything at all, it just opens a PowerShell and closes in half a second.

I believe that switching to 64-bit should definitely fix my previous issues with memory.  But, I was wondering a few things :

•  Should I just try to compile the source code? (I have never done anything like this).
•  Did I miss something in the above-mentioned processes?
•  Does structure 64-bit also need to be in the same directory as param files and source dataset? (As 32-bit needs to)
•  Is it possible for you to send the compiled 64-bit structure?

Thank You

 Vaibhav

[Francisco Pina Martins]

Dear Vaibhav,

> I realized the issue, I noticed that in my system, the structure runs
> are still running with the 32-bit version.

So that’s why! Makes perfect sense.

> I expected that after installing structure_threader via pip/pip3, a
> 64-bit version of the structure would be compiled/stored/installed?
> but it is not. as mentioned by you in the answers

> <https://groups.google.com/g/structure-software/c/SsOXY9Y8t8o/m/Kwn0o5SCCwAJ>to

> one of the questions before in this group. or maybe I am not able to
> find it? maybe it does not download 64-bit in Windows?

I haven’t used windows as a daily driver OS in a *very* long time (as in
almost 20 years now). I wouldn’t even know where to start to compile a
STRUCTURE binary from source code on that platform. /Structure_threader/
is meant for HPC workloads, and therefore the provided binaries are
built and tested on GNU/Linux (OSX binaries were created by a colleague
at the time and haven’t been tested in years).

> Also, I am not able to download the 64-bit structure directly via the
> website

> <https://web.stanford.edu/group/pritchardlab/structure_software/release_versions/v2.3.4/html/GWA.html>,

> the link doesn't work. But Again you have given the updated link

> <https://web.stanford.edu/group/pritchardlab/structure_software/release_versions/v2.3.4/structure_kernel_source.tar.gz>

> in the group already for the source code

> <https://gitlab.com/StuntsPT/Structure_threader/-/blob/master/helper_scripts/install_structure.sh>

> where I could compile it.

Indeed there does not seem to be a binary available for it on the
official website. Just 32bt windows builds.

> Along with the source code

> <https://web.stanford.edu/group/pritchardlab/structure_software/release_versions/v2.3.4/structure_kernel_source.tar.gz> of

> the 64-bit structure,  you have also provided a helper script

> <https://gitlab.com/StuntsPT/Structure_threader/-/blob/master/helper_scripts/install_structure.sh> of

> structure_threader, which can download and compile the structure. Does
> this script also work on Windows? I tried to run the script, but it
> does not seem to do anything at all, it just opens a PowerShell and
> closes in half a second.

The helper script is for GNU/Linux systems. It will certainly not work
in Windows.

To answer your final questions:

*

Yes, you will have to compile it from source. I have no idea how to
do that on Windows, but maybe someone else in this forum does (but
there are other platforms more suited for that than this group)

*

I don’t think you missed anything. Your assessment seems good.

*

The way 64bit STRUCTURE handles PATHS should be the same as the
32bit version.

*

I have no way to build or provide a 64bit windows binary file

The way I see it, here are your options:

1.

Use a smaller (subsampled) dataset

2.

Find a way to get a 64bit STRUCTURE binary (most likely to succeed
is to compille the source code yourself)

3.

Use a GNU/Linux based OS to perform the analysis

4.

Use a compatibility layer such as WSL <https://ubuntu.com/wsl> to
run your analyses

5.

Run your analysis under a Docker container

Hope at least one of these can help you.

Best,

Francisco

On 02/02/24 09:16, Vaibhav Mahilang wrote:

> Dear Francisco,

> Thank You.
>
> I realized the issue, I noticed that in my system, the structure runs
> are still running with the 32-bit version.
>
> I expected that after installing structure_threader via pip/pip3, a
> 64-bit version of the structure would be compiled/stored/installed?
> but it is not. as mentioned by you in the answers

> <https://groups.google.com/g/structure-software/c/SsOXY9Y8t8o/m/Kwn0o5SCCwAJ>to

> one of the questions before in this group. or maybe I am not able to
> find it? maybe it does not download 64-bit in Windows?
>
> Also, I am not able to download the 64-bit structure directly via the
> website

> <https://web.stanford.edu/group/pritchardlab/structure_software/release_versions/v2.3.4/html/GWA.html>,

> the link doesn't work. But Again you have given the updated link

> <https://web.stanford.edu/group/pritchardlab/structure_software/release_versions/v2.3.4/structure_kernel_source.tar.gz>

> in the group already for the source code

> <https://gitlab.com/StuntsPT/Structure_threader/-/blob/master/helper_scripts/install_structure.sh>

> where I could compile it.
>
> Along with the source code

> <https://web.stanford.edu/group/pritchardlab/structure_software/release_versions/v2.3.4/structure_kernel_source.tar.gz> of

> the 64-bit structure,  you have also provided a helper script

> <https://gitlab.com/StuntsPT/Structure_threader/-/blob/master/helper_scripts/install_structure.sh> of

> structure_threader, which can download and compile the structure. Does
> this script also work on Windows? I tried to run the script, but it
> does not seem to do anything at all, it just opens a PowerShell and
> closes in half a second.
>
> I believe that switching to 64-bit should definitely fix my previous
> issues with memory.  But, I was wondering a few things :
>

> *  Should I just try to compile the source code? (I have never done
> anything like this).
> *  Did I miss something in the above-mentioned processes?
> *  Does structure 64-bit also need to be in the same directory as

> param files and source dataset? (As 32-bit needs to)

> *  Is it possible for you to send the compiled 64-bit structure?

> /Structure_threader/ will automatically create one for you to make

>> I had never seen /Structure_threader/ fail only on

>> specific replicates of a repeated analysis…
>>
>> From the screenshots it seems you were able to perform
>> all runs simultaneously on your machine. K18 replicates
>> are likely to have been the last to finish. Since they
>> errored out and no log was written I have only one
>> suspicion. Perhaps you have run out of storage space on
>> the machine? If you are 100% sure you did not, can you
>> please share the command you used to run

>> /Structrure_threader/?

>>
>>> Also, can you explain more about, How only STRUCTURE can
>>> be able to make a single run use multi-core? I tried to
>>> use only STRUCTURE and it is using only a single core.
>>
>> I am not aware of any way to do this. All the wrappers I
>> know (strauto, parallelstructure) do the same as

>> /Structure_threader/ - make each separate and independent

>>> <https://web.stanford.edu/group/pritchardlab/structure_software/release_versions/v2.3.4/structure_doc.pdf>.
>>> The |-R| option in /Structure_threader/ refers to

>>> how many runs you want to perform for each value of
>>> |K| (you need multiple runs to estimate the “best”
>>> |K| using Evanno’s method).
>>>
>>>> 2 See the picture attached SNAP.png, when using
>>>> replicates -R = 1 and K's = 16,17 and 18, the
>>>> structure threader only runs in a single thread for
>>>> each 16,17,18 K's for 10000 NUMREPS. so in total
>>>> only using 3 threads. What do I do so that the
>>>> threader can use all the threads available?
>>>

>>> /Structure_threader/ is a wrapper program for

>>> STRUCTURE. As such it cannot make single runs work
>>> with multiple cores (only STRUCTURE would be able to
>>> do that). What you can do to leverage your multicore
>>> system is run more values of |K| and/or add
>>> replicates with |-R|.
>>>
>>>> 3 I understand that each line shown in the Python
>>>> console (in picture SNAP.png),  is running 10000
>>>> NUMREPS in the background (Meaning that the
>>>> structure threader is using structure in the
>>>> background ). Is it in any way possible to see each
>>>> NUMREPS in the Python console, as we were able to
>>>> see via the structure program? ( attached picture
>>>> structure.png)
>>>
>>> Although you can’t look at the NUMREPS information
>>> “live” as in STRUCTURE, you can run

>>> /Structure_threader/ with |--log 1| option which

>>>> I have released /Structure_threader/ version

>>>> /Structure_threader/. Let's try to

>>>> get you up and running. =-)
>>>> From what you show, I suspect that
>>>> this is a combination of parameters
>>>> that will cause STRUCTURE to
>>>> produce output files slightly
>>>> different from what

>>>> /Structure_threader/ expects to

>>>> find. If this is so, I'd like to
>>>> get it fixed as soon as possible.
>>>> In order to speed testing, can I
>>>> please request that you post the
>>>> exact command you used to run

>>>> /Structure_threader/, as well as

>>>> your `mainparams` and `extraparams`
>>>> files, please? I will attempt to
>>>> replicate your run locally and
>>>> figure out exactly where the issue
>>>> lies.
>>>> Thank you!
>>>>
>>>> Francisco
>>>>
>>>> A quarta-feira, 20 de dezembro de
>>>> 2023 à(s) 15:11:18 UTC, Vaibhav
>>>> Mahilang escreveu:
>>>>
>>>> Dear Developers and Everyone,
>>>>

>>>> I am using *structure *(Windows

>>>> Command Line Version) for my
>>>> analysis.
>>>>
>>>> Currently, I have Windows 11, 8
>>>> GB RAM. Which I used for a few
>>>> trial runs with the expected
>>>> population (K) based on
>>>> real-life assumption, 1000
>>>> burn-in, and 1000 repetition.
>>>> Everything worked very well.
>>>> Results are also aligned with
>>>> the expected outcome.
>>>>
>>>> Now, for actual analysis, I
>>>> have access to a server PC with
>>>> Windows 10, 128 GB RAM, and 16
>>>> threads, which is available
>>>> solely for this specific
>>>> analysis. As expected I am

>>>> using *structure threader* for

>>>> using the capabilities of
>>>> server pc.
>>>>
>>>> Since I have switched to

>>>> *Structure Threader* I have

>>>> started receiving a few errors,
>>>> which I haven't gotten to fix
>>>> by myself. The parameters to
>>>> run the analysis and errors are
>>>> as follows :
>>>>

>>>> *Parameters - Structure Threader*
>>>>
>>>> *Mode -* run
>>>> *Klist -* 17 18 19 (Real life
>>>> population number - 18)
>>>> *Repetitions each K -* 10
>>>> *Threads -* 16
>>>> *Log -* 1
>>>>
>>>> *Structure mainparams*
>>>>
>>>> *MAXPOPS -* 18
>>>> *NUMIND -* 424
>>>> *NUMLOCI -* 68436
>>>> *BURNIN + NUMREPS -* 100 (Why

>>>> Low? It is a trial run)

>>>> *POPDATA -* 1
>>>> *POPFLAG -* 1
>>>>
>>>> *Structure extraparams*
>>>> *NOADMIX -* 0
>>>> *USEPOPINFO -* 1
>>>> *INFERALPHA -* 1
>>>> *RANDOMIZE -* 0 (Structure

>>>> Threder sets it to 0 by itself )
>>>>

>>>> *ERRORS*
>>>>
>>>> *1 *Structure harvester cannot

>>>> read lnPr(X|K) Estimated Ln
>>>> Prob of Data

>>>> *2 *Evanno method cannot be
>>>> applied
>>>> *3 *Cannot generate plots
>>>> *
>>>> *

>>>> To figure out the issue I tried
>>>> to run the structure threader
>>>> with different parameters and
>>>> also tried to check the
>>>> results. Here are my observations :

>>>> *
>>>> *
>>>> *- *Results do not contain

>>>> Estimated Ln Prob of Data
>>>> (Explains Error 1 and 2)
>>>>

>>>> *- Tried (No test for best K)
>>>> *still cannot generate plots

>>>> - Error  - value error for
>>>> plotting (See attached picture

>>>> - *NoKTest.png*)
>>>> - *Tried ( No plots) *

>>>> - Error 1 and 2 (as expected)
>>>> (See attached picture -

>>>> *NoPlotTest.png*)
>>>> - *Tried (No test for best K +
>>>> No plots) *

>>>> - Logfile,f, and q file
>>>> produced for every repeat
>>>> - Plots have to be generated
>>>> manually
>>>>
>>>> My question is did I miss
>>>> something that caused the ln
>>>> probability to not generate?
>>>> What should I do to fix this?

>>>> *
>>>> *

>>>> I apologize for the long
>>>> thread, I wanted to give a
>>>> clear idea of my analysis.
>>>> Any help is appreciated.
>>>>
>>>> --
>>>> You received this message because you are
>>>> subscribed to a topic in the Google Groups
>>>> "structure-software" group.
>>>> To unsubscribe from this topic, visit
>>>> https://groups.google.com/d/topic/structure-software/E_zevPCM-a8/unsubscribe.
>>>> To unsubscribe from this group and all its topics,
>>>> send an email to structure-softw...@googlegroups.com.
>>>> To view this discussion on the web visit
>>>> https://groups.google.com/d/msgid/structure-software/981cc395-32e0-429e-9a94-99eb0f833825n%40googlegroups.com

>>>> <https://groups.google.com/d/msgid/structure-software/981cc395-32e0-429e-9a94-99eb0f833825n%40googlegroups.com?utm_medium=email&utm_source=footer>.

>>> ​
>>>
>>> --
>>> You received this message because you are subscribed to
>>> a topic in the Google Groups "structure-software" group.
>>> To unsubscribe from this topic, visit
>>> https://groups.google.com/d/topic/structure-software/E_zevPCM-a8/unsubscribe.
>>> To unsubscribe from this group and all its topics, send
>>> an email to structure-softw...@googlegroups.com.
>>> To view this discussion on the web visit

>>> https://groups.google.com/d/msgid/structure-software/759f6127-b81d-42ff-9646-260b9dc02595n%40googlegroups.com
>>> <https://groups.google.com/d/msgid/structure-software/759f6127-b81d-42ff-9646-260b9dc02595n%40googlegroups.com?utm_medium=email&utm_source=footer>.

>> ​
>> --
>>
>> You received this message because you are subscribed to
>> the Google Groups "structure-software" group.
>> To unsubscribe from this group and stop receiving emails
>> from it, send an email to
>> structure-softw...@googlegroups.com.
>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/structure-software/048f08e7-d581-4e01-a98b-b8ee1f62e1dc%40gmail.com

>> <https://groups.google.com/d/msgid/structure-software/048f08e7-d581-4e01-a98b-b8ee1f62e1dc%40gmail.com?utm_medium=email&utm_source=footer>.

>>
>> --
>> You received this message because you are subscribed to a topic
>> in the Google Groups "structure-software" group.
>> To unsubscribe from this topic, visit
>> https://groups.google.com/d/topic/structure-software/E_zevPCM-a8/unsubscribe.
>> To unsubscribe from this group and all its topics, send an email
>> to structure-softw...@googlegroups.com.
>> To view this discussion on the web visit

>> https://groups.google.com/d/msgid/structure-software/a8fcf726-d99b-4427-97dd-dade67fb11bcn%40googlegroups.com
>> <https://groups.google.com/d/msgid/structure-software/a8fcf726-d99b-4427-97dd-dade67fb11bcn%40googlegroups.com?utm_medium=email&utm_source=footer>.

> ​
>
> --
> You received this message because you are subscribed to a topic in the
> Google Groups "structure-software" group.
> To unsubscribe from this topic, visit
> https://groups.google.com/d/topic/structure-software/E_zevPCM-a8/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> structure-softw...@googlegroups.com.
> To view this discussion on the web visit

> https://groups.google.com/d/msgid/structure-software/ddd9ea94-c33b-4987-9b32-b79675590604n%40googlegroups.com
> <https://groups.google.com/d/msgid/structure-software/ddd9ea94-c33b-4987-9b32-b79675590604n%40googlegroups.com?utm_medium=email&utm_source=footer>.

​