Size doesn’t always matter

Size doesn’t always matter

The autosomal genome of Ötzi the Austrian “Iceman” is apparently in the pipeline (from what I can tell they’re doing the analysis right now). What can we learn from one sample? Ann Stone, who was a graduate student on the original team which recovered his body, says:

A specialist in anthropological genetics, Stone is excited by the recent news but also cautious. “It is a sample of one. For us to really say something about that period, you need a sample of 25 to 50 individuals,” she explained during an interview with Deutsche Welle, Germany’s international broadcaster.

This is fine as it goes. Worries about sample size are pretty generic and if the practicalities permitted who wouldn’t want a bigger N? But whether you should worry about sample size is partly conditional on how much the findings deviate from what you’d expect. Imagine for example that ~25% of Ötzi’s genome was of Neandertal origin. Obviously it would be great to have 25 to 50 representative individuals from this region to know whether Ötzi was atypical…but the very finding itself would be of such large effect that an N = 1 would tell us quite a bit. Similarly, one genome of a Sub-Saharan African would be very informative if you had several hundred non-African genomes as a point of comparison (because Sub-Saharan Africans have so much genetic variation which is outside of the distribution found among non-Africans).

Razib Khan