lauantai 7. marraskuuta 2015

Introducing Treemix

During the next days or weeks I am going to run several Treemix-analyses targeting to regional migrations among present-day Europeans.  My first run figures migrations in Finland, but please keep on following this work, because I'll do several analyses covering most regions in Europe and neighboring areas.

What we see in Finland (as derived from the result)

- prehistoric Germanic migration to Southwest Finland
- Swedish migration to Finland
- post-Tavastian migraton to Estonia
- Finno-Ugric (pre-Finnic) migration to Komi (which definitely indicates Komi being a mixed Finno-Ugric population with European root of another kind

Post-Tavastian likely  means the age before Finland was populated by present-day Finns.

There is also a migration from France to Poland, but solving it needs more western and Central European samples.  Next runs will be more focused on this issue.

This Treemix run was executed using global setting, with no root and the migration count 5.

6 kommenttia:

  1. The migration edge from France to Poland probably just indicates LBK admixture in Poles. We should not take the admixture edges too literally. Also TreeMix can be of some help but is not any exact science tool.

    Something that makes me wonder in your graph is the general topology of the tree. I find most intriguing, perplexing even, that the Scandinavian+Western branch and the Finnic one appear as "siblings", while most mainline Eastern Europeans, including Poles, Russians and Estonians, appear upstream it the tree topology, and Latvians and Lithuanians even more upstream, what feels very unnatural in contrast with other data. This makes me suspect that the choice of samples and/or the particular tree (what do less migration edges produce?) is not the best.

    I mean: why on Earth would Finnish and French cluster at all? Why do Estonians and Baltics appear near the root? It does not make any sense to me. Maybe try adding Basques and Sardinians for improved topology?

  2. Being downstream in the tree just indicates drift away from the upstream, it doesn't mean the downstream branches are necessarily closest to each other. Lack of migration edges would make that tree look very different, no doubt.

    1. Yeah, it looks like being drift. But I am just in the beginning in this work and want to see more results before conclusions. But, as I mentioned in another answer, changes in tree shape when running without defined root population seems to have little (none?) effect on migration routes.

    2. But what about Baltic Finnic admixture in Scandinavia, or at least in Sweden? Various analyses indicate that it exists, TreeMix should demonstrate that as well in some way. Can it do so by showing Finnic migration to Scandinavia, or does it do so by showing the opposite?

    3. It is possible. Sure the Baltic Finns and Swedes have some common old ancestry. However those Swedish samples usedn in this test are very pure Swedes and do not show Finnish admixture. I have tested it and I am sure of it. I have several other Swedish samples and see real Finnish admixture among them, but not among those academic samples. So what we can expect here in this test is a common old ancestry between Baltic Finnic and Swedish people. Is it notable on this graphics, I don't know, Maybe. The picture is not simple; there is common old Fennoscandinavian ancestry, there is old Germanic ancestry in Finland and there is more recent common ancestry on both sides.

  3. Thanks for your observations, they really need explanation. I use here no-root setting, so there is no root population to make any continuation from the past. In this run Treemix assumes that most homogeneous populations (Balts) are the root. If I run this several times I see that the tree topology changes and Balts and Slavs are not alway in the root. But this (changes in the topology) has very little effect on migrations, there are still same migrations to Finnish populations even though the tree shape changes. But why I do not use root population? Because there are only two possible solutions to use it, neither good one. First I could use very distant root, like chimp or Mbuti, but this doesn't serve my attempt to find out Bronze or Iron Age migrations. Treemix tends to give highest ranks for migrations routes with highest genetic distance, so using Chimp or Mbuti gives Native American, Siberian or similar very old migrations. On the other hand I could select a closer root to solve this problem, but it is not easy to find common root for all tested populations, being appropriate and acceptable for all tested populations. So I didn't use root definition at all and as a resuilt the tree shape varies, but as I mentioned it doesn't effect on found migrations. If there are differences in migrations routes between Treemix runs it seems to depend on the used migration algorithm.