Tag Archives: genetic distance

A principal component analysis of human populations

12 Wednesday Jun 2019

Posted by in Uncategorized

≈ 53 Comments

Tags

, , , , ,

Principal Component Analysis (PCA) is a mathematical technique by which many data points get reduced to a smaller number of more manageable data points.

Cavali-Sforza lumped humans into nine major populations. The following shows his phylogenetic tree of these nine populations followed by a matrix showing the genetic distance between them:

Because I wanted to see if these nine populations could be objectively reduced to a smaller number, I made all the distances negative and then entered the genetic distance matrix into a minitab spread sheet.

The reason I made the distances negative is because PC analysis is usually done on correlation matrices where the higher the value, the more similar. In a genetic distance matrix, it’s the opposite, hence the negative signs I added.

The principal component analysis gave the following result.

C1=Africans, C2= Non-European Caucasoid, C3=European Caucasoid, C4= Northeast Asian, C5=Arctic Northeast Asian, C6= Amerindians, C7= Southeast Asian, C8= Pacific Islander, C9= New Guinean & Australian

To determine how many principal components to retain, mathematicians use what’s called the eigenvalue > 1 rule, which in this case means only three components.

The first component explains 54% of genetic variation and since Northeast Asians have the highest loading on this component (0.432), it can be thought of as a measure of Northeast Asianness. Africans are the only group to load negatively on Northeast Asianness (-0.376).

The second component explains 26% of the variation and since Europeans have the highest loading on this (0.526), it can be considered a measure of whiteness.

The third component explains 12% of the variation and since Native Americans have the highest score on this (0.527) it can perhaps be considered a measure of “New Worldliness”.

Now when I plot each of the nine populations in three-dimensional space (x axis = Northeast Asianness, y axis = whiteness, z axis = New Worldliness) with their loadings multiplied by 10 to make differences visible, we find all of the nine populations fit into three major clusters.

made using Graphing Calculator 3D

These three clusters are extremely similar to the three major races of physical anthropology: Mongoloids on the back wall, Negroids on the side wall, and Caucasoids on the floor.

No disrespect to Caucasoids (I’m 100% pure Caucasoid myself). The graph can be reoriented so any group is on the floor.

One anomaly is that New Guineans & Australian aboriginals cluster with Mongoloids, even though they are morphologically closest to Negroid. Of course such anomolies are not uncommon in taxonomy. Birds for example genetically cluster with reptiles, even though they’re not reptiles. Humans cluster with apes, even though we’re not apes.

Such anomalies occur because most of our DNA is junk, so it groups us based on how recently we share common ancestors, not by how much of that common ancestor we shared.

Genetic distance between humans, neanderthals & other primates

25 Saturday May 2019

Posted by in Uncategorized

≈ 92 Comments

Tags

, , , , , , , ,

About a year ago I wrote:

The following chart (created by some scientist(s) led by David Reich) shows the genetic divergence between hominin samples as a fraction of the human-chimp difference.  So for example, all the human groups have just over a 0.12 genetic divergence with Neanderthals, meaning that the genetic difference between humans and Neanderthals is only 12% as great as the genetic difference between humans and Chimps (source: supplement of Genetic history of an archaic hominin group from Denisova Cave in Siberia.)

distance4

The purpose of the chart is to estimate how long ago the different populations diverged from a common ancestor.  So since the fossil record tells us that Neanderthals and chimps diverged about 6.5 million years, then humans and Neanderthals should have diverged roughly 0.8 million years ago (12% of 6.5 million) assuming genetic divergence maps to chronological divergence in a linear way.

I transformed the genetic distance matrix into a dendrogram, which looks at all the distances and creates the most parsimonious family tree:

Dendogram showing genetic relationship between 8 groups. From left to right: chimps, san, Yoruba, French, Papuan, East Asian, Denisovan, Neanderthals

What’s cool about dendrograms is they let you determine the number of categories and subcategories in a very objective way.

Of course dendrograms are only as good as the data you put into them, and I don’t endorse basing taxonomy simply on genetic relatedness, but if I did, here’s how I’d interpret the above tree:

The first major split is between chimps & everyone else. This is consistent with two well recognized genera of hominins : Pan (i.e. chimps) and Homo (humans and near-humans).

Now within the Homo genus, we see another major split in the tree. Anatomically Modern Humans (AMH) vs Archaic Humans. Thus we can divide the homo genus into at least two major species.

Within the Archaic Humans we can further subdivide into major races: Denisovans and Neanderthals.

Now within our own species, AMH, the dendrogram shows three major races: Capoids, Congoids and Non-Africans.

I’m not saying I agree with this taxonomy since it was only based on genetic distance (much of which is junk DNA) but what’s great about using dendrograms is almost everyone looking at them will assign groups to the same categories and subcategories, even if they don’t use the same words (race, species, genus) to describe them. It’s wholly objective.

But what is needed is a dendrogram based on polygenic scores of actual phenotypes. That way people who have the same phenotypes caused by the same genomic architecture could be grouped together.

Unlike the above dendrogram, which groups based on how recently we share a common ancestor, we need to group based on how much of the common ancestor we share.

[update may 26, 2019: an earlier version of this article misspelled dendrogram]

[2nd update may 26, 2019: an earlier version of this article contained bragging that has since been removed]