Here is a link to my DIY Eurasia K20 Calculator https://drive.google.com/file/d/0B5L...ew?usp=sharing. I have also included a populations averages csv file that can be used with 4 Oracle. I have tested both the calculator and oracle and verified they work.
My calculator is based on around 2400 samples from Haak, Allentoft, and other public datasets. Please post genotype rate an other calculator output data along with your oracles.
A couple of things to keep in mind:
1- This is a high K calculator, so it is expected that there will be a calculator effect, which I will be able to judge from your oracles. This should be mitigated a little by the fact that none of my signals are closer than around 0.08 FST from each other.
2- The steppe signal is quite narrow, and centered around Yamnaya. This is partly due to the ratio of my Yamnaya to non-Yamnaya steppe samples being on the high side. As a result your overlap would be less than if the Steppe signal was broader and centered around more recent ancient samples. Due to Yamnaya being an older sample, it is expected that the post-Yamnaya drift of modern populations would be greater than say with a BA Sintashta genome.
3- You will notice that Caucasus/Gedrosia is a single component. This is due to the fact that Caucasus and Gedrosia are not very differentiated signals, and I did not take extra measures to attempt to separate them. I may do so in a future calculator.
4- These results are based on an unsupervised run, with solid cluster for the most part. I wanted to leave the output natural.