Well, for anyone interested in using the Damgard et al samples, you might want to consider this comment posted by Kurd:

"Notice to anyone using the Damgaard samples or other aDNA samples. There are many samples with coverage issues, and many which are complete outliers to the populations they are identified with.

Whether you are a blogger, researcher, or hobbyist, it’s IMPERATIVE that you take a couple of days to identify and remove those problem samples the datasets, otherwise analysis can get screwed (results not making sense) up whether using f3s, f4s, qpAdm, or ADMIXTURE, by those problem samples.

I have myself spent the past few days doing just that, and have removed dozens of samples with the goal of having the most accurate dataset out there BEFORE starting analysis.


Edit: qpAdm is extremely tricky to use (Patterson will be the 1st to tell you that). Results greatly vary based on the selection of rt pops. Even I’m not very comfortable in using it. Thus no one should rely solely on qpAdm to reach any conclusions. F3s and f4s are much safer to use and base conclusions on."