Re:Aboriginal Australian genomes reveal Indian ancestry
by ImaCake on Wednesday March 01, 2023 @08:22PM
(#34990736)
> Is an 11% genetic incursion a lot for ~ 150 generations?
I am not sure either. However, we do have a lot of natural experiments (and lab experiments) to explore this.
The 11% number comes from a clustering analysis. The user supplies k = number of ancestry groups, and the algorithm estimates proportion of genetics specific to each of k ancestry groups. So basically fancy k-means clustering. I don't know how robust it is and I would suspect a biased estimate.
As for over time... also complex. But I would imagine the best models would use exponetial distributions or similar as their basis (i.e. initially fast decrease in percentage followed by a slower rate of loss until equilibrium is reached).
I am not sure either. However, we do have a lot of natural experiments (and lab experiments) to explore this.
The 11% number comes from a clustering analysis. The user supplies k = number of ancestry groups, and the algorithm estimates proportion of genetics specific to each of k ancestry groups. So basically fancy k-means clustering. I don't know how robust it is and I would suspect a biased estimate.
As for over time... also complex. But I would imagine the best models would use exponetial distributions or similar as their basis (i.e. initially fast decrease in percentage followed by a slower rate of loss until equilibrium is reached).