I will fix this, should be quick.

Hello.

I am trying to run Structure_threader to use the multi-core capacity put I am stock on this problem.

I have tried solving the path problem with C:\Program~2\ and "c:\programfiles (x86)..." but neither worked as I still get the same error that Structure cannot be found in the specified path...

Do you have a workaround to this issue?

I would really be happy if I could use your solution has right now I have 7 idle cores...

Thanks!
Pedro

We really should make them quit with an error status.

Just what the title describes.
To further document the issue, the output files of structure become somewhat different when the mentioned flag is set.

This is what we expect under "normal" circumstances:

Inferred ancestry of individuals:
        Label (%Miss) Pop:  Inferred clusters
  1  Coc_W.1    (0)    1 :  0.033 0.967 
  2 Coc_W.10    (0)    1 :  0.032 0.968 
  3 Coc_W.11    (0)    1 :  0.033 0.967 
  4 Coc_W.16    (0)    1 :  0.033 0.967 
...

And this is what we get when "USEPOPINFO" == 1:

Inferred ancestry of individuals:
Probability of being from assumed population | prob of other pops
        Label (%Miss) Pop
  1 neoNAAri1    (0)    1 :  0.999 | Pop 2: 0.000 0.000 0.001  |  
  2 neoNAAri2    (0)    1 :  1.000 | Pop 2: 0.000 0.000 0.000  |  
  3 neoNAAri3    (0)    1 :  1.000 | Pop 2: 0.000 0.000 0.000  |  
  4 neoNAAri4    (0)    1 :  1.000 | Pop 2: 0.000 0.000 0.000  |  
...

So not only e have to deal with a different number of lines, as we also have to deal with a different organization of the clustering...

Thanks to @JulianBaur for triggering and reporting this.

Currently, when using Structure_theader for plotting, an argument -i is required to set a prefix for the Q-values files.
However, this is actually only a prefix and requires the user to be in the directory where the Q-values files are located, which is less than ideal.
We should fix this to allow passing a full path to the argument.

All the cool kids are doing it! xD

The logging builtin should be use to output information into the terminal. Some of the advantages would be a standardized way to provide messages to users, control over the verbosity of the logging, custom error messages, etc.

We must find a way to graciously hanlde failiures (which are not errors, just lack of conditions to perform tests). Using quit() instead of sys.exit() seems to be insufficient.

Using a popfile will create the divisions as requested, but the bars themselves are not changing places.
The final results is the plot being drawn by the input file order, but the population labels and divisions are being given to the wrong bars.

When the --params switch is provided for STRUCTURE.
The program should issue a warning that we are running without an extraparams file, but carry on.

When dealing with OS paths, the "/" directory separator should not be hardcoded, since this will not work in Windows. Instead, we should use os.path.join() that will handle OS differences for us.

This should simplify the whole process for the user.

Traceback (most recent call last):
  File "bin/structure_threader", line 11, in <module>
    sys.exit(main())
  File "/home/qsuber/.local/lib/python3.5/site-packages/structure_threader/structure_threader.py", line 568, in main
    create_plts(arg.outpath, wrapped_prog, Ks, bestk, arg)
UnboundLocalError: local variable 'bestk' referenced before assignment

Need to investigate.

The link that is created when wrapping fastSturcture is made with the full link path. This will break if the files are moved to another location, or if the analysis is performed inside a container.
Steps need to be taken to ensure the created link simply points to the target file on the same dir, without the full path.

Currently, only STDOUT is caught and logged.

For added convenience.

As per the title. I will have to look at the randrange function and understand what is going on.

Fix in progress.
Currently blocked by #56

Title says is all.
Currently if an unexisting file is specified as --pop argument, plotting will fail with a python error. We don't want that.

If a user provides a structure file that is not sorted by population, the result is a messy plot.
The plot function (or even better, the parser) should sort the individuals by their population value before plotting.

As per the title.

I'd like to run your tool with a fastStructure wrapper, but my input files are plink formatted. Do you have a method for converting these to fastStructure format? Or does your tool currently work with plink formatted files?

Has some issue with using an already existing dir.

It will lead to cleaner code and more functions can be added later on with minimal hassle.

Also allows being used with pip.

Title says it all. At the end of the run the evanno test should automatically be applied to select the ideal "K" value.

faststructure (https://github.com/rajanil/fastStructure) should also be supported in our threading program. This will not add much in terms of speed gains, but should provide some much needed convenience.

Yeah, windows again.

Pretty much what's in the title.
Needs an argument option for this. Would be cool if it could be conditional to -fs.