Ethiopia Pop Analysis

## Last updated: Mon Feb 03 13:35:00 2014

Summary of Data

plot of chunk unnamed-chunk-3

PCA Scatter plot by defined populations

plot of chunk unnamed-chunk-4

The populations defined by location do not show any observable clustering. Therefore, the samples will be clustered by k-means based on first 3 pricipal components.

Below, we see PCA scatter plots coloured by K-means clustering on first 3 PCs

plot of chunk unnamed-chunk-5 plot of chunk unnamed-chunk-5

Above, both figures show PC1 vs PC2 but coloured by k-means groups 2 and 3. In the second plot, some points seem to overlap, but they are seperated along the z-axis. Now we try to plot PC with elevation using the k-means groups.

PC vs elevation scatter plots coloured by k-means clustering on first 3 PCs

plot of chunk unnamed-chunk-6 plot of chunk unnamed-chunk-6 plot of chunk unnamed-chunk-6

Above, we see no clustering based on elevation. In fact, all groups are at all elevations.

Now we try find.clusters from Adegenet to group individuals.

plot of chunk unnamed-chunk-8 plot of chunk unnamed-chunk-8

We get similar results to k-means, but still no correlation to elevation.

Spatial Analysis and sPCA

Spatial Locations

plot of chunk unnamed-chunk-9

Here, PCA PC1 is plotted to spatial coordinates.

plot of chunk unnamed-chunk-10

From above, PCA principal components show no spatial structure.

Now we carry out an sPCA analysis. Seen below are sPCA eigenvalues and screeplot.

plot of chunk unnamed-chunk-11 plot of chunk unnamed-chunk-11

Below is shown sPCA Global Score 1 actual scores and lagged scores.

plot of chunk unnamed-chunk-12

Below is shown sPCA Global Score 2 actual scores and lagged scores.

plot of chunk unnamed-chunk-13

Ultimately, these samples do not show any population structure using STRUCTURE or PCA or sPCA. The samples are not correlated to elevation by location grouping or by k-means clustering.

End of Document.