The Principal Component Analyses (PCA) below are based on my K8 model (aka. fateful triangle) and ancient genomes from Haak et al. 2015. A spreadsheet with the K8 ancestry proportions for the ancient samples is available here. Some of the results are a bit noisy, and there are good reasons for that (like low coverage calls), but overall I think they look quite solid.
Note the total lack of the Ancient North Eurasian (ANE) component in the Middle Neolithic (MN) genomes, and its sudden appearance at levels of around 24% in the Late Neolithic (LN) Corded Ware genomes. Keep in mind that these samples are from the same region of Germany and only separated by a couple thousand years at most. Clearly, what we're seeing here is a major migration to Central Europe from the east.
So how do I know that these K8 ancestry proportions are correct? Because when I analyze several of the highest quality genomes with very different methodology, like genotype-based PCA, I get basically the same outcomes.
Bell_Beaker I0112 PCA
Corded_Ware I0103 PCA
Karelia_HG I0061 PCA
Yamnaya I0231 PCA
Yamnaya I0443 PCA
However, I have to admit that I'm now more befuddled than ever as to why anyone would want to model the Yamnaya as 50/50 Karelia_HG/present-day Armenian. I do realize that Haak et al. showed this to be a solid statistical fit, but I just don't see it as a very practical solution considering the surprisingly high ANE and low Near Eastern ancestry in the Yamnaya genomes.
I suspect that inflated Basal Eurasian admixture in the Yamnaya and/or East Eurasian, perhaps ancient Arctic, admixture in Karelia_HG might have skewed the Haak et al. model.
Haak et al., Massive migration from the steppe was a source for Indo-European languages in Europe, Nature, Advance online publication, doi:10.1038/nature14317
4mix: four-way mixture modeling in R
Modeling Yamnaya with qpAdm
K8 results for selected Allentoft et al. genomes