fastStructure endless "fail" returned

[Lorenzo Bertola]

Hello,

I am using fastStructure on a mac. I managed to run both structure.py and chooseK.py on my full dataset (i.e. all individuals). The results made sense, they matched what I obtained with DAPC and STRUCTURE. So up to here no problems.

I do not think the issue resides here, but just to be sure:

- macOS High Sierra 10.13.4

- running in conda virtual environment

- python 2.7

- cython 0.27 <- later versions returned errors

Using chooseK.py, I identify K = 3 as the configuration that best matches my data.

Then, as often routinely done nowadays, I tried to re-run fastStructure on each individual cluster, to try and identify any possible finer structure.

I produced the subsetted input file in 2 ways:

- extracted each group with adegenet::seppop within R, and then converted using radiator::genomic_converter. This is the way I produced the original working file including all individuals

- I manually split the original working file in bbedit, retaining only the individuals belonging to each K

Both times, structure.py starts running and produces a log file (see content below), but instead of producing output files it starts saying:

failed

failed

failed

failed

...

I waited a few minutes and it keeps doing this with no sign of stopping. From the log file it seems the run had ended after a few seconds (this cluster has only 15 individuals).

Here the content of the log file:

Marginal likelihood with initialization (1) = -1.0948722684

Marginal likelihood with initialization (2) = -1.0948830764

Marginal likelihood with initialization (3) = -1.0948692217

Marginal likelihood with initialization (4) = -1.0948406796

Marginal likelihood with initialization (5) = -1.0948721821

Iteration Marginal_Likelihood delta_Marginal_Likelihood Iteration_Time (secs)

0 -1.0948406796 -- 0.039

10 -0.9787746076 0.1160660720 0.074

20 -0.9787672747 0.0000073330 0.049

30 -0.9787671949 0.0000000798 0.051

Marginal Likelihood = -0.9787671949

Total time = 0.2155 seconds

Total iterations = 30

Thanks for any help,

Lorenzo

[Vikram Chhatre]

The marginal likelihood values across the runs within a cluster are very close to each other.  This may be why the CV fails.  When you remove CV flag, do you get meanQ output? If so, are you seeing any significant split of ancestry coefficients between 2 or more clusters?

--
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/ed0c6e0f-7348-4300-a2bd-eec8076540f0%40googlegroups.com.