Net Nucleotide Distance and Expected Heterozygosity

[Leah]

I am getting some very strange results for allele frequency divergence
(net nucleotide distance) and average distances between individuals in
the same cluster (expected heterozygosity). My expected
heterozygosity is coming out to be negative values like -151 and net
nucleotide distance is 20 and above.
Has anyone had this problem and figured out a way to fix it?
LC

[Luc]

I had the same problem.
The formulae are in the software documentation pp 25 and 24
http://pritch.bsd.uchicago.edu/software/structure_v.2.3.1/documentation.pdf
Note that there is a mistake in formula (5)!
in the rightmost term, you should read (HA+HB) not (HA-HB)

I made the calculations using a spreadsheet (tedious but not really
difficult) based on the allele frequencies in the bottom part of the
text output.

The "net nucleotide distance" given in the Structure output was
proportional to the true value.
For expected heterozygosity, there is a linear relationship between
the printed result and the true value, which is calculated based on
the usual formula (6) in the doccumentation)

I will try to write to the authors.

Luc

[Pall Melsted]

Hi

I am the new programmer maintaining structure. I've looked into the
issue and can't recreate the bug on the data I've tested. I've
verified this using reported allele frequencies and formulas (5) and
(6), (with the correction of H_A + H_B, which is the way it is done in
the program).

If it is possible to send the datasets, I could recreate the bug and
hopefully fix it and get it out as part of the next release.

Best regards,
Pall Melsted

On Dec 18, 4:41 am, Luc <luc.baudo...@cirad.fr> wrote:
> I had the same problem.

> The formulae are in the software documentation pp 25 and 24http://pritch.bsd.uchicago.edu/software/structure_v.2.3.1/documentati...

[Leah]

How can I e-mail this data to you?
LC

[Pall Melsted]

Hi Leah

You can send the data file and the parameter files if you have them to
pmel...@gmail.com

thanks,
Pall Melsted