I have observed differences between results obtained using Admixtools installed on different servers, one with Ubuntu 18, one with Centos 7. This is occurring specifically when I am performing outgroup f3 statistics using qp3Pop and I narrowed it down to the version of qp3Pop used
These are the type of differences I obtain for f3
Version 412 or 420 600 or 650
Test1 0.178593 53.963371
Test2 0.180800 54.504648
This is not linked to the computer but to the version of the software qp3Pop. 412 was compiled on the first computer, 420, 600 and 650 were conda installed on the second computer, but 650 was also conda installed on the first computer.
What are the underlying causes of these differences? Are both types of results sounds (there seems to be some kind of constant relative behaviours but the values cannot obviously be compared when different versions of qp3Pop are used)?
Thanks for the clarification.
Thierry
I have observed differences between results obtained using Admixtools installed on different servers, one with Ubuntu 18, one with Centos 7. This is occurring specifically when I am performing outgroup f3 statistics using qp3Pop and I narrowed it down to the version of qp3Pop used These are the type of differences I obtain for f3 Version 412 or 420 600 or 650 Test1 0.178593 53.963371 Test2 0.180800 54.504648 This is not linked to the computer but to the version of the software qp3Pop. 412 was compiled on the first computer, 420, 600 and 650 were conda installed on the second computer, but 650 was also conda installed on the first computer. What are the underlying causes of these differences? Are both types of results sounds (there seems to be some kind of constant relative behaviours but the values cannot obviously be compared when different versions of qp3Pop are used)? Thanks for the clarification. Thierry