Totalt antall sidevisninger

tirsdag 28. august 2012

New analysis of the ancient gotlanders vs modern European populations

EDIT 30/08-2013: I now consider this analysis outdated. Please check the more recent posts with reanalysis of Ajv52, Ajv70, Ire8, Gok4 and Ste7.

Its time for a new ancient gotlander analysis using 22k, this time using the same reference panel and individuals as in the 289k analysis. However as 22k SNP's is a rather small number of SNP's the resolutation is much smaller and we will see that populations blend more into each other. In addition the ancient gotlander also affect the analysis because of these individuals unique genetic variation.

We will first look at the individual chunkcount aggregated heatmap. Much is familiar from previous analysis. We see generally that the modern populations to less extent paint there chromosome with the ancient gotlanders than the ancient gotlander paint the chromosome with the modern populations. This is a sign that the ancient gotlanders and modern populations have very different haplotypes because of divergence in time and difference in genetics. If we included a single San individual from Africa in the analysis we would probably see the same.

In the PCA plots we see that the ancient gotlanders place themself in dimension 1-2 in between the Vologda Russians to the left, Lithuanians to the right, the Estonians to the upper right. This seems to indicate that the geographical close populations of Lithuanians and Vologda Russians.

However if we look at the ancient gotlanders position in dimension 1-3 we find the ancient gotlanders close to part of the Finns to the left and close below to one Saami SWE7 and a few Scandinavian-Saami mixed individuals. This make it difficult to explain the ancient gotlanders only in the context of Lithuanians and Vologda Russians. Also in lower dimensions the ancient gotlander is in its own space seperate from the modern populations suggesting it has variance not seen in any current populations.

As we can see the plots are much more fuzzy than in the analysis we have seen earlier using 289k SNP's so it is very possible that we see large variations at the individual level. If we average the number of chunkcounts we see a clearer picture of affiliation. We see here that the Saamis appear to show the largest affiliation on average as both donor and receiver to the ancient gotlanders.

If we try to see these distributions geographically we appear to see that the ancient gotlander appear to affiliate with the more remote parts of the continent while the more central continent with the exception of the French appear to be the most different from the ancient gotlanders.

We also see a strange and a very different affiliation between the Finnish groups. West-Finns appears to affilaite well while the East-Finns appears to affilaite very little even when close populations like Estonians, Vologda Russians and Saamis appear to show affiliation. At the time beeing I cant explain this difference.

As the map suggests below maybe there have been a later genetic expansion to the north from maybe more south-weas European contries that have changed the genetic landscape since the time of the ancient gotlanders especially to the south-south-east.

Other maps and plots

Distinctive haplotypes in Saamis and Finns

I want to return to the differences one sees between donors and receivers. The sharp observer would probably notice the difference between the donation and receivers between f.ex Finns and the Chuvash or even in general between Saamis and Finns and other populations. The Finns dont use much Chuvash chromosomes to paint there chromosomes but Chuvash use on the other hand more Finn chromosomes to paint their chromosomes.

In the "newick" format it could be written as C(F1,F2). It has nothing to do about ancestral to who but is a tree representing three current populations. Where C vs F1,F2 coalesce at a time t1 and F1,F2 coalesce at a time t2. As we know it is correct to assume that F1,F2 coalesce at an later time than vs the C and probably F1,F2 similar coalesce time to C.

If we then assume that F1,F2 have have very distinctive haplotypes the painting of F1,F2 will ignore copying chunkcounts from C as it can find closer haplotypes internally, but C on the other hand is equally distant to F1,F2 will copy smaller chunkcounts from both of these. This will result in a symmetrical chunkcount heatmap.

ID    C    F1    F2
C     0    0.5    0.5
F1   0     0      1
F2   0     1      0

This can be seen very clearly as an extreme case with the ancient gotlanders vs modern populations, where the modern populations copy very little chunks from the ancient gotlander but the ancient gotlander copy much chunks from the modern populations as there a no other to copy from. If we had than one ancient gotlanders the picture would of course change. It would probably also be similar if we included a single San individual from Africa.

So asymmetrical donation-receiving for a population can be explained by distinct haplotypes for that population vs other populations.

(Thanks to the authors of Finestructure for their explanations and patience!)

søndag 26. august 2012

Saami dimension in the PCA

Saami and Finns have always been close to each other in the higher dimensions of the PCA plots and heatmaps from the previous analysis, however in dimension  D1-D6 the Saamis clearly diverge away from the others in the panel into their own dimension. Finns here appears here to go in a complete opposite direction sharing space with Vologda Russians.

We see very clearly in this dimension the individuals of main Saami ancestry cluster into their own dimension while Finnish and Scandinavian individuals with Saami minority ancestry pull straight towards the individuals of mainly Saami ancestry.

This likely means that Saami do have distinctive haplotype segments that seperate them clearly from the other populations including Finns.

onsdag 15. august 2012

Scandinavians - Western Europeans with minor Saami-Finnish admixture

Changes from the last analysis: Added Bulgarians. Used complete reference popualtions. New participant FI23. FI4 have opted out. Else same specifications as in earlier runs. Adding more individuals increase the accuracy of the software but the increased diversity may smooth out smaller differences.

As we can see here from the previous run is that the main PCA dimensions have switched from the last run and we see the main structure we have often seen in the standard unlinked SNP MDS plots run through Plink software.

PCA plots shows independent genetic variations. The first dimensions shows the largest variations while the second dimensions shows the second largest variation and so on. A indvidual with equal background from two different populations would typically appear in line in the middle between these populations, so would also be the case for recently mixed populations. If foundereffect and genetic drift have occured in a population the population would typically not be in line between populations, however on PCA plots its also quite possible that the limited number of panel populations can give a similar effect on the PCA plot. In this PCA plot we are looking at the number of chunks (genetic segments including identical and related segments that have mutated).

As commentet in earlier post it appears like in the PCA plot D1-D2 (D1 horizontal, D2 vertical) that Scandinavians pull away from the other European populations because of Finnish-Saami like admixture early in its founding as they as a group all have it in various minor degree. It explain its in-line location between British and French and the Finns-Saami. It can be confirmed in the ChunkCount heatmaps (see results section) where Scandinavians shows strongest connections to these populations, but do not show same influence to other continental European populations. In experiments mixing events give lower number of chunkcounts to parents populations than the parents have to their own populations and within the new mixed population. We see this in Scandinavians, West-Finns and mixed groups like the two Bothnian groups. More recent mixed individuals vs Finns and Saamis is to be found further upstream towards the Finn and Saami.

(PCA plot D1-D2 with population assignment)

We also see this in the PCA plot D1-D3 (D1 horizontal, D3 vertcal). However the combination with the D3 dimension make to difference the influences on the Scandinavians. We see that the upper half of the Scandinavian cluster is totally dominated by Swedes. It appears like Swedes may have got more influence from Finns while the Norwegians on the other hand appears be more influenced by the Saamis. The pull upwards may be because of a weaker relationship to Lithuanians according to the heatmap.

We further see that Finns in D3 appear to have an more intermediate position between the Saamis and the Vologda Russian, Mordovians and Estonians. In the heatmaps however we only see increased influence towards the Vologda Russians and Estonians, not the Mordovians. This indicate that the pull upward in D3 is due to increased relationship to the Vologda Russians and Estonians. Note that the two Finns FI23 and FI5 have confirmed Saami minority admixture so the relationship appears to be real.

In the heatmaps for the Saami we also see a increased relationship to the Vologda Russians however this influene may be a from a common source not directly from the Vologda Russians but from a third common Chuvash like source. This seems also indicated by the PCA plots as the Saami appears to pull more towards the Chuvash not the Vologda Russians..
(PCA plot D1-D3 with population assignment)

fredag 3. august 2012

New Finestructure Analysis

Its time for a new Finestructure analysis. The difference from the earlier Finestructure analysis are more project participants and more reference populations.


Lets first look at the chunkcounts matrix, that is the number of shared segments including both identical and related segments that have mutated (closest neighbouring segment). In the matrix the rows are receiptants and columns are the donors.

Two main branches Saami-Finns and Continental-European-Scandinavian

We see the formation of two main branches: 1) Common Continental European and Scandinavian branch 2) Saami-Finnish branch. The former branch divide again into two subbranches a Western European branch and a Eastern European branch. The Scandinavians appears to divide at the earliest branch from the Western European branch. Early branching indicate more distant relateness, late branching indicates closer relatedness. In the eastern European branch the Chuvash break away the earliest from the rest and it make sense as they are a distant population. In the latter Saami-Finnish branch the Bothnian group appears to divide the earliest from the Saami and Finns and it make sense as they likely have more Scandinavian like influence. The arrangment of trees vs other trees is arbitrary so you need to look at the heatmap and branch level to make any judgment about relationship.

 (ChunkCount aggregated for individuals)
(ChunkCount aggregated with population labels) 

Bothnians - two groups with different influences

In the heatmap above we see the Bothnian2 group who shows strong relationship to Scandinavains have higher sharing with the Saami. Finns do not show a similar affiliation to this group. The group share especially with the Saami2 group that have that is known to have at least partly South-Saami ancestry. Several individuals in this group also have documented Saami ancestry from the South-Saami areas. This suggest that the reason for the Bothnian2 groups seperation from other Scandinavians is due to South-Saami influence. We also see this in the PCA plot below where the Bothnian2 group close to the Scandinavians appears to pull straight towards the Saami and not to Finns.

The Bothnian1 group appears to have a somewhat different history than the Bothnian2 group, while the latter appears to be Scandinavains with Saami admixture the former appears to be more like Finns with Scandinavian admixture. We see this both in the heatmaps and in the PCA plots where there is a pull toward the Scandinavians. There also appears like this group in general appears closer to Continental-European populations than the Finns.

Finns - two groups with more western and more eastern influences.

Internally Finns divide into two groups East-Finns Finn1 and West-Finns Finn2. Eastern Finns appears to have an increased sharing between each others compared to West-Finns this probably to a late expansion into eastern and northern Finland. Finns appears to affiliate to in decreasing order Bothnian1, Bothnian2, Estonians, Scandinavians and Vologda Russians. These are all geographically close or even overlapping groups. In the finer grades we observe that Eastern Finns appears to affiliate less with Scandinavians then Western Finns. East-Finns also appears affiliate less with more western continental populations than western Finns. Western Finns seem also to be a little closer to Lithuanians than western Finns but the difference is small.

We see from the PCA plot below that Eastern Finns and Saami dominate the far right of the first dimension (horizontal) but all Finns are above the horizontal line in the second dimension (vertical). This may indicate a common genetic backgound for Saami and East-Finn in that dimension. We do not know the "original" position for East-Finns but if they originally where closer to the Saamis position on the vertical their higher position on the vertical may be due to Baltic influences, however we also see that East-Finns have the same vertical height as the Vologda Russians so it is possible that there is a direct connection between Vologda Russians and East-Finns f.ex that the Vologda Russians originally had a larger Finnsih like component that later become more similar to slavs due to centuries of migration and assimilation moving Vologda Russians to its current position.

We see further that Western Finns appears right in line between Eastern Finns and Scandinavians. This indicate that Western Finns have more Scandinavians influence also seen in the heatmap and it is the reason why Western Finns appears closer to Scandinavians. This admixture may be one factor why West-Finns appears more diverse.

We also see in the PCA plot that Estonians, Mordovians and especially the Vologda Russians appears to pull away from the Lithuanians, Belorussians and Ukrainians towards the Finns. This indicate that even these populations are continental they have a significat Finnish like components.

(ChunkCount aggregated PCA with population labels)

Saami - various degree of Scandinavian and Chuvash like influence?

The Saami divide into two groups. The Saami1 group consists of Saami entirely of North-Saami origins while the Saami2 group also contain at least partly individuals of South-Saami origin. The heat map suggest a slightly closer connection to East-Finns than to West-Finns. The Saami2 groups shows increased relationship to the Bothnian2 group than the Saami1 group. The Saami1 group shows increased affinity to the Vologda Russians and Chuvash. Both groups show same affinity to the Estonians. The Saami2 group also shows increased affinity to the Scandinavians. To sum it Saami appears to seems to affiliate with more eastern population and strongest to its closest neighbours except for the Vologda Russians and Chuvash.

The closer position of Saami2 in the PCA plot on the direct line to the Scandinavian may suggest Scandinavian influence however we do not know why East-Finns and the North-Saami group cluster to the far right. It may be the result of foundereffect and/or drift as no ancestral population can be inferred from this axis alone except it is shared by East-Finns.

On the vertical axis on the other hand we see that the Saami is at the same level as Italian, French, British and lower part of the Scandinavian cluster. It may indicate that Saami have ancestral relationship to these populations, however the positioning of the Chuvash and the increased chunk sharing indicate that the Saami maybe have been influenced vertically by a population with elements similar to the Chuvash or Vologda Russians. It may be a reflection of common Siberian like ancestry deteced in earlier analysis or an artificat of the PCA software as the Chuvash appears quite distant to the Saami in other dimensions. Further in earlier analysis there appear to be a strong correlation between Koryak, Chuvash and Vologda Russians in the Fennoscandia dataset.

Scandinavians - western, eastern and northern influences

The Scandinavians all fall into one group. Its part of the western/central european branch where it divide from the other subgroups at the earliest branch. In the heatmaps we see they are closest to the Botnian2 group that divide into a subgroup because of minority Saami ancestry. Next they affiliate to a subgroup of Brits. Then they also cluster to the Bothnian1 group that appears to be a group close to the Western Finn group pulling towards the Scandinavian cluster. Of the more distant populations the Scandinavians then cluster to the French. Then also to Finns especially the Western Finns. Scandinavians also shows increased affinity to Mordovians, Belorussians, Lithuanians and Ukrainians, actually more than Finns and Saami in general. However they have lower affinity to the Vologda Russians and Chuvash than Finns and Saami.

In the PCA plot we see that Scandinavians pull slightly away from the continental clusters towards partly the Finns and partly the Saami where they are much in line with the Belorussian-Ukrainian and the Lithuanian cluster. One may here suggest that the reason for this positioning is that Lithuanians and the Belorussian-Ukrainians do have a common backgroun in dimension but a check of the heatmap suggest that the highest intensity is between the British-French and the Finns-Saami with the former having the highest intensity. So the pull of Scandinavians from the French and the British seems to be the result of Finns and Saami influence, on the other hand one cant exclude that French and British have been pulled away from the Scandinavisn either by influenes not seen in this population panel.

On the vertical we see that the Scandinavians appears to be slighty moved upwards compared to the French and the British. This is possibly due to the pull from the Lithuanians and Belorussian-Ukrainians as Scandinavians appears to show a stronger pull to these populations in the heatmaps than to Romanians and Hungarians. If we look at the individual PCA plot we can also see that Swedes appears to dominate the upper right part while Norwegians appears to dominate the lower left part. This reflect that Swedes appears to have more eastern influences both in the direction of Lithuanians and Belorussians-Ukrainians and in the direction of Finns. The Norwegian on the other hand tend to move towards the western populations like French and the British suggesting they have more western like influences.

(ChunkCount aggregated PCA D1-D2 with individual labels)
(ChunkCount aggregated PCA D1-D3 with individual labels)

(ChunkLenght aggregated with population labels)

(ChunkLenght aggregated for individuals)

(ChunkLenght aggregated PCA D1-D2 with population labels)
(Unable to seperate the coloring for one Saami group)

(SegmentSize CL/CC aggregated for individals)
(SegmentSize CL/CC aggregated with population labels)
(SegmentSize CL/CC aggregated PCA with population labels)

(SegmentSize CL/CC aggregated PCA with individual labels)

(Updated 13 Aug 2012)

onsdag 1. august 2012

MDS Update 85 individuals

Its time for a new MDS update.

We see much the same structure in the two first dimensions D1-D2 as in earlier analysis. The X or horizontal axis D1 shows south-west Europe at the far left. The variance then as we go further right into two branches on the Y-asis or D2 where the upper branch contains continental European populations while the Scandinavians, Finns and the Saamis form the lower branch. At the edge of the continental branch we find the Chuvashes while on the edge of the lower branch we find the Saamis and Finns.

The second plot shows D1-D3. Here we see on the vertical D3 Lithuanians on the top and the Italians and the Chuvash at the bottom. Finns and Scandinavians appears to cluster abit over the middle while the Saamis appears to cluster abit lower than the middle at the same levels as French, Romanians and Italians.

Third plot D2-D3 is a combination of D2 and D3 explained above.