Nm=H???

[Jean]

Hi Joao,

In the target files and the simulated data files (.dat), the value of
Heterozygosity (H) always equals that of Nm (for a given population).
Unless I'm mistaken, this shouldn't be.

Here an example of simulated data (2 populations, microsat marker):
[top|marker|avMutM|sdMutM|avRecM|sdRecM|t1|Ne1|Ne2|NeA1|mig1|mig2]
5 summstats [H1|H2|NmH1|NmH2|H1-2]

1 1 -5.5 0.4 0 0 106611 202826 93140.4 3.86261e+06 0.0570107 0.0877647
0.943089 0.922298 0.943089 0.922298 0.92815
1 1 -5.5 0.4 0 0 106546 876724 354867 9.90696e+06 0.0844052 0.072692
0.953542 0.949658 0.953542 0.949658 0.950664
1 1 -5.5 0.4 0 0 107269 663084 343305 936894 0.0667429 0.00983277
0.371661 0.310534 0.371661 0.310534 0.329847
1 1 -5.5 0.4 0 0 106047 337529 552649 4.84192e+06 0.0526817 0.0529092
0.974448 0.968536 0.974448 0.968536 0.969119
1 1 -5.5 0.4 0 0 102323 107755 989729 3.04631e+06 0.0559678 0.00682203
0.81417 0.792066 0.81417 0.792066 0.801427
1 1 -5.5 0.4 0 0 104594 845411 943057 4.63755e+06 0.0980893 0.00154784
0.944251 0.928865 0.944251 0.928865 0.935531

Could you look into it?
Also, what's the meaning of "H1-2"? Apparently it's not |H1-H2|.


Thanks.

Jean

[Joao Sollari Lopes]

Hi Jean,

Nm is calculated as Hw/(1+Ha-Hw), where Hw is the heterozygosity
measured within a population and Ha is the heterozygosity of all the
populations pooled together.

It is expected, then, that Hw < Ha. When this is not the case Nm = Hw.

This is done so that the value of Nm increases up to Hw as the
difference between Ha and Hw decreases. As soon as Hw >= Ha, the value
of Nm = Hw.

H1 is the heterozygosity in population 1

H2 is the heterozygosity in population 2

H1-2 is the heterozygosity of the population 1 and 2 pooled together
(which tends to be higher than H1 and H2 when there is little
migration between the populations 1 and 2)

Best,
Joao

[Jean]

Thanks for you reply, but I'm still missing something.

When there are 2 populations, Ha = H1-2, right?
In the simulations posted here, H2 < H1-2 (= Ha).
But we still have NmH2 = H2 in all cases.


Jean


On Feb 16, 4:30 pm, Joao Sollari Lopes <j.sollari.lo...@gmail.com>
wrote:

[Joao Sollari Lopes]

Hi Jean,

You're absolutely correct.

There is a bug in the code when calculating NmH. More specifically,
when calculating heterozigosity for all the populations pooled
together (Ha) the software is set to skip all the microsat loci
instead of sequence loci. This results on a Ha of zero. So everytime
NmH is calculated Ha is always going to be smaller than Hw, which
means that the NmH of a population is always going to be set to its H.

This means that you shouldn't use NmH. Deeply sorry about that.

Thank you very much for your contribution! I'll make sure to correct
that for the next release!

Best,
Joao

[Jean]

Thanks for investigating, Joao.

I'd like to use popABC to study migration rates. Without NmH, it's
going to be harder. But I assume I can compute NmH (or any relevant
index of gene flow) based on H1, H2 and H1-2 (since I study only 2
populations). It will just require some R script on the .dat file (and
target file too), before doing the regression step.

Best,

Jean

On Feb 17, 12:23 pm, Joao Sollari Lopes <j.sollari.lo...@gmail.com>