I noticed a few days ago that Admixture had an update available:

1.1 (2/8/2011): Parallel processing, supervised analysis. Minor speedups and cleanups.

There were two important new features in version 1.1 that I started salivating over. One was parallel processing so I could utilize all the cores of my machine and thus run Admixture faster. The other was more important though I have yet to experiment with it. It's the ability to assign some ancestral components to specific samples, i.e. assign some individuals in the data specific 100% ancestry as a starting assumption and calculate admixture from that.

Of course, these two features made me forget the cardinal rule: Never upgrade in the middle of an analysis. But I did upgrade and things have changed subtly, making some comparisons between admixture v1.04 and v1.1 difficult.

For example, previously (admixture v1.04), at K=12, admixture was giving me the ancestral components: South Asian, Balochistan/Caucasus, Kalash, Southeast Asian, Southwest Asian, European, Papuan, Northeast Asian, Siberian, East African Bantus, West African, and East African.

With Admixture v1.1, I am getting the ancestral components: South Asian, Balochistan/Caucasus, Kalash, Southeast Asian, European, Mediterranean (maximum among Mozabite and Sardinians), Papuan, Northeast Asian, Southwest Asian, Siberian, West African, and East African.

So now I am running Admixture with different random seeds and trying to compare the old version results vs the new. Of course since we are talking K=12, just one admixture run takes a whole day.

Anyway, while that's going on, I have more things in process which can go forward, like reporting the results of Batch 4. And working on the Eurasian dataset.