Hi Danila,

If I understand correctly, I agree. if all you did is switch the P1 and P2 labels, then your ABBA and BABA counts should simply be inverted. What tool did you use for the calculation?

Hi @Simon H Martin,

It seems I figured out my problem. I used CompD which allows performing of multiple tests simultaneously and I obtained both ABBA and BABA patterns depending on used samples. As I understand, when negative value of D and positive values of Z-score show prevalence of BABA pattern, then opposite case- positive D and negative Z-scores tell us about prevalence of ABBA pattern.

One additional question. Following authors recommendations, Z-score more than 3 or 4 allows us to assume significant introgression. What if I obtained z-scores values between 2 and 3 in all tests with different samples used as P1 and P2? May I gently presume existence of at least minor gene flow?

Regarding your additional question:

You cannot assume minor geneflow from z-scores. When you want to know the extend of the introgression you can use the f3-test. The z-score gives you an idea of how likely it is that your D is indeed different from 0 and not just a result of randomness. If you know the underlying distribution, you can calculate a p-value from your z-score. Assuming a normal distribution, a z-score above 2 should actually be enough to be somewhat sure that the result is significantly different from 0 (a z-score of 2 corresponds to a p-value of ~0.045 in a two tailed test). I am not quite sure why the authors recommend 3 or 4. Where did you read this?

Yes Timo Hellwig

Thank you, Timo Hellwig. I got it and will try to perform f-test. Concerning z-score > 3 . As I understand, it corresponds to significance values more than three standard errors different from zero.