Calculation of DeltaK as in Evanno´s paper

[Marian]

Hi! I´m doing my analysis in STRUCTURE and I would to know how can I
calculate the deltaK value as in the Evanno´s paper.Is there a formule
or something to facilitate the calculation? I also do a graphic LnP(D)
vs K and the best k was 4 but I think k=2 is best for my population.
Also when I saw the structure graphics, with K=2, I have 90% of
individuals with Q>0.7, with k=3 I have 82,5% of individuals with
Q>0.7 and with k=4 I have 67,5% of individuals with Q>0.7.
How should I justificate my selection of k?
Thanks in advance!
Mariana

[Vikram Chhatre]

Hi Mariana,

An easy way of estimating delta K is using Structure Harvester program
(google it). It has some specific requirements.

1. Equal number of runs per consecutive value of K: If you're testing
5 clusters, you should do say 10 runs (or however many you want, but
keep it uniform) for K1, K2, K3, K4 and K5. In other words you can't
do k1, k2, k4, k5 or k1, k3, k5.

2. Once all runs have finished, create a zip archive containing all of
the results_f files and use that as Structure harvester input.

3. Harvester will output a tarball archive containing your results
which can be visualized in the browser as well as downloaded for your
records.

Let us know if you need any further help.

Vikram

> --
> You received this message because you are subscribed to the Google Groups "structure-software" group.
> To post to this group, send email to structure...@googlegroups.com.
> To unsubscribe from this group, send email to structure-softw...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/structure-software?hl=en.
>
>

[Gabriel Eduardo Melim Ferreira]

Hi Mariana, there's a great tool for this: Structure Harvester. Very easy and helpful.
http://taylor0.biology.ucla.edu/structureHarvester/index.php

regards
Gabriel

2011/9/15 Marian <marianf...@gmail.com>

--
You received this message because you are subscribed to the Google Groups "structure-software" group.
To post to this group, send email to structure...@googlegroups.com.
To unsubscribe from this group, send email to structure-softw...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/structure-software?hl=en.

--
Gabriel Eduardo Melim Ferreira

[Marian]

Thank you very much to both of you! It really works this program!
With Ln(k) vs k I have K=4.
With the Evanno method I have K=3
And with DeltaK vs K I have =3, this last two is not the same?
So I think I should use K=3 correct? Can I also justify my selection
also because the percent of individuals with Q>0.7 with K=3 was 82.5%
and with K=4 was 67,5%?
Thanks!
Mariana


iel
>
> 2011/9/15 Marian <marianfasane...@gmail.com>

[AjaY..]

Dear Community
how to give no. of iterations in Linux machine. i had K- 1 to 4. I did not find options in main parameters and extraparameters files..