Totalt antall sidevisninger

torsdag 19. desember 2013

Updated Fennoscandia Analysis

(Updated 26/12-13) Its finally time for updated Fennoscandia analysis. Its been over a half year since the last update. We have more than 53 new participants most of these are Swedes.

We do as usually look first at the heatmap that catches the main structure of the data. CC is ChunkCounts and is the total number of shared segments including identical and related segments. CL is ChunkLenght and is the total shared (both identical and related) segments in cM (centimorgans).

CC Linked Aggegated 289k 
 CC Linked Raw 289k  

CL Linked Aggregated 289k

CL Linked Raw 289k

The most different we see from previous Fennoscandia level analysis is the huge influx of Swedes creating a bigger cluster especially for the northern part (Norrländingar) of the country.  This cluster of Swedes at the middle of the heatmap appears to show affiliation to the Southern Saami cluster consisting of SWE7 and SA3. So it appears like there must have been some kind of founder effect and maybe drift in this process that have made them look more related to each other than other Swedes and other Scandinavians from further south. This foundereffect appear to have included a Saami population or Saami like population. 

There is also a secondary cluster consisting of mostly Swedes but also some Norwegians in the upper part of the heatmap below the Saami. These appears to show affiliation to the Northern Swedish cluster, Finns and individuals of mixed Scandinavian and Finnish ancestry.

External influenced varuation

The PCA plots can as seen before be divided into external influenced variation and internal variation. Dimension 1 and 2 appear to be very similar external variations that peaks in "Others" and with Danish and Scandinavians and lowest among Finns. That this variation peaks among Danish suggest a central-European origin of this external influence. These two dimension effectively divide Scandinavians and Finns.

 CC Linked PCA D1-D2 289k 

Dimension 3 on the other hand reflect an external variation from "Others" that peaks among Finns and to large extent Saami and the least among Scandinavians suggesting a Eastern European continental origin for this influence. As we can see also this dimension appear to be effective to differentiate Finns and Saamis from Scandinavians but it doesnt appear so clear cut as dimension 1 and 2 above.

 CC Linked PCA D1-D3 289k 

Internal influenced variation

As dimension 1 horizontal axis appears to differentiate so well the Scandinavians and Finns at each extreme we will continue to use this dimension. All the dimensions above 3 appear to show internal variation. The dimension 4 vertical axis appears to show the Scandinavian-Finn and Saami at each extreme.

CC Linked PCA D1-D4 289k

We clearly see from the new participants that NO25, SWE56, SWE62, SWE53 and SWE52 very clearly pull out of the Scandinavian cluster towards the Saamis and to some lesser degree also SWE59 and SWE55. Most of the appear to belong to the " Norrländingar" cluster.

Of individuals going away from the Finnish cluster towards the Scandinavians we can mention FI35, FI40. Of Scandinavians moving towards Finns we can mention NO21, SWE65, SWE54, NO26, SWE45, SWE55, SWE59, SWE50, SWE68.

As we can see from the heatmap the north-Swedish cluster appear to be much closer related to each than the normal Swede. This also shows in the PCA plot dimension 1-5 with dimension 1 at the horizontal and dimension 5 at the vertical.

 CC Linked PCA D1-D5 289k  

The "Norrländing" cluster appears to peek in Norrbotten area. As we can see the South-Saamis (SWE7+SA3) appears to attract the most of other groups to this cluster.

DImension D1-D6 appears diffcult to explain. It appears to be Finnish variation as its more dispersed among Finns than others.

CC Linked PCA D1-D6 289k

I will need some more research about this dimension. It doesnt seem to reflect the structure of Finnish subgroups in the heatmaps.

CC Linked PCA D1-D7 289k

In this plot the horizontal dimension 1 is the same as above while the dimension 7 appears to peak among the South-Saami individuals SWE7 and SA3. As we can see that both some Finns and Scandinavians climb towards these individuals. However the top other individuals closest to these South-Saamis appears to be from Västergotland and Skania area. On the Finnish side I am not sure but maybe coastal areas of Finland or from Karelia.

This clustering make me unsure and it maybe will not be cleared out properly with higher marker resolution, something I have been planning to do at least for the Fennoscandia analysis to higher power of differentiation.

I have also been looking at dimension 6 and it doesn't make sense to me currently so it add further uncertainty to the interpretation of dimension 7 a higher dimension than dimension 6 as higher dimensions means more uncertainty as individual variation become more visible in the data than in the lower dimension that capture more at the group level. 

(Work in progress)

1 kommentar:

  1. Interesting results, Anders. As I was expecting, Sa1 seems to be connected to Eastern Finns such as myself, not the other way around. The other Saami don't behave in the same way in the plots. As being an outlier in many plots as Fi14, I wonder which might be the currently missing populations I'm apparently attracted to?