Negative outgroup f3 statistics
sallycylau opened this issue · comments
Hi!
Thank you for creating this program.
I have performed an outgroup f3 statistics test between 4 populations of my study species (popA - D), with the outgroup species I used being a very closely related species with interspecific gene flow observed in areas where their distribution overlap.
For my outgroup f3 statistics test, I have deliberately chosen to only analyse populations and outgroup samples that are not from areas where they overlap, with no evidence interspecific admixture detected between them (based on Structure, PCA).
However, the returned f3 values are negative.
| A | B | C | f3 | stderr | Zscore | nsnps |
|---|---|---|---|---|---|---|
| popA | popA | outgroup | 0 | -1 | 0 | -1 |
| popA | popB | outgroup | -0.086 | 0.00187 | -46.1 | 22839 |
| popA | popC | outgroup | -0.0866 | 0.00187 | -46.4 | 22197 |
| popA | popD | outgroup | -0.0834 | 0.00178 | -46.9 | 18162 |
| popB | popA | outgroup | -0.086 | 0.00187 | -46.1 | 22839 |
| popB | popB | outgroup | 0 | -1 | 0 | -1 |
| popB | popC | outgroup | -0.0867 | 0.00191 | -45.4 | 24651 |
| popB | popD | outgroup | -0.0843 | 0.0018 | -46.7 | 22538 |
From my understanding, negative f3 happen when C is admixed with A and B, or that C is just admixed, which makes biological sense to me. Even though I have negative f3 values, I wonder if I could still infer the the shared genetic drift between popA and B etc with respect to my outgroup? Or will the principle of outgroup f3 statistic collapse once I get negative values?
Thank you very much in advance.
Best regards
Sally