Vahaduo

[Siegmund]

Hey there!

What is the difference between Distance and Single on Vahaduo? Sometimes there I've got totally different outcome.

[AimSmall]

Individuals vs Populations/Groups

[ltnultihaxer]

What is considered good distances for modern vs ancient model and what is considered a bad distance?

[TanTin]

What is considered good distances for modern vs ancient model and what is considered a bad distance?

I would suggest to avoid using such methodology. Because modern are heterozygous, and ancient are more homozygous.  So if you want to compare them it depends how the software used is dealing with heterozygote.

[ltnultihaxer]

What is considered good distances for modern vs ancient model and what is considered a bad distance?

I would suggest to avoid using such methodology. Because modern are heterozygous, and ancient are more homozygous.  So if you want to compare them it depends how the software used is dealing with heterozygote.

Oh I didn't mean comparing a modern vs a ancient model, what i meant to say is when running modern models what is considered a good distance and when running ancient models which is considered a good distance. 

Just curious to know what is seemed good and bad for each type of model but not necessarily comparing them!

[Genetics189291]

What is considered good distances for modern vs ancient model and what is considered a bad distance?

I would suggest to avoid using such methodology. Because modern are heterozygous, and ancient are more homozygous.  So if you want to compare them it depends how the software used is dealing with heterozygote.

Oh I didn't mean comparing a modern vs a ancient model, what i meant to say is when running modern models what is considered a good distance and when running ancient models which is considered a good distance. 

Just curious to know what is seemed good and bad for each type of model but not necessarily comparing them!

Anything reaching close to 1 percent or below is considered a overfit, you need to be very careful of the sources you are using. You can alternatively just reduce to 3 or 4 populations to get a basic break down

[ltnultihaxer]

What is considered good distances for modern vs ancient model and what is considered a bad distance?

I would suggest to avoid using such methodology. Because modern are heterozygous, and ancient are more homozygous.  So if you want to compare them it depends how the software used is dealing with heterozygote.

Oh I didn't mean comparing a modern vs a ancient model, what i meant to say is when running modern models what is considered a good distance and when running ancient models which is considered a good distance. 

Just curious to know what is seemed good and bad for each type of model but not necessarily comparing them!

Anything reaching close to 1 percent or below is considered a overfit, you need to be very careful of the sources you are using. You can alternatively just reduce to 3 or 4 populations to get a basic break down

Thank you so close to 1 is an overfit, what would be considered an underfit? Something above 2 or 3?

[Kale]

When dealing with ancients at least, there are varying degrees of noise, or when using G25 data, lack of a sufficiently proximate source, meaning 1% might be overfit for one sample, and others might not get 1% no matter what.
I think a good methodology for ancients is to...
1) No limit on the sources, let it overfit, and see what the distance is.
2) Start restricting the number of populations incrementally. If the distance increases just a little each time, that's perfectly ok.
3) Stop when the distance jumps substantially.