Tuesday, September 27, 2016

ANI, ASI, etc.

CIP wrote in the comments to the previous post about the 2016 Reich paper:
This is in accord with the conventional view that Europeans and Asians probably separated after leaving Africa in the Middle East. East Asians then separated from South Asians in India and Amerindians and related groups separated from East Asians much later.
Guest wrote in the comments about the 2009 Reich paper:
One, the ‘Ancestral North Indians’ (ANI), is genetically close to Middle Easterners, Central Asians, and Europeans, whereas the other, the ‘Ancestral South Indians’ (ASI), is as distinct from ANI and East Asians as they are from each other. By introducing methods that can estimate ancestry without accurate ancestral populations, we show that ANI ancestry ranges from 39–71% in most Indian groups, and is higher in traditionally upper caste and Indo-European speakers. Groups with only ASI ancestry may no longer exist in mainland India. However, the indigenous Andaman Islanders are unique in being ASI-related groups without ANI ancestry.
Guest wrote this in another comment:
You are confusing the ANI and Indians who currently live in the North. All Indians studied in the Reich paper (except Andamese) are mixtures of ANI and ASI, and consequently more related to each other than to outside groups like West Eurasians. The two papers are quite consistent, and David Reich is an author on both papers.

My response:



The 2009 paper has a diagram like this (my diagram is not to scale, but is topologically correct):

What does the 2016 paper tell us?  It tells us that all Indians (all the mixtures of ANI and ASI) are neighbor-joining closer to a group that includes East Asians and Amerindians; and in line with the CIP comment above, the chart should look like (topologically, but not to scale):

East Asian and Amerindian form a group.  While per 2016, all mixtures of ANI and ASI are closer to this group than to Europeans, I'm not moving ASI from its position in the 2009 paper.  

I think it is remarkable that the 2016 paper seems to state that
1. All mixtures of ANI + ASI are closer to the (East Asian + Amerindian) group than to European.

2. India has definitely had mixing with peoples from its west, whereas India has been reproductively isolated from the Amerindians for more than 14,000 years.  Despite this, Indians (all ANI, ASI mixtures) are closer to (East Asian, Amerindian).   It seems that the mixing simply does not show up on the resolution of the 2016 chart.
Remember, the neighbor-joining algorithm creates a group of all (ANI, ASI) mixtures.   It creates a group of (East Asians, Amerindians).  It creates a group of various {Europeans}.  And it then chooses to group the (ANI, ASI) mixtures with the (East Asian, Amerindians) group, and not with the {Europeans}  group. Note that in the groups mentioned in the 2016 paper,  Pathans are around 71% ANI, Madiga around 32% ANI.
3.  Certainly, the topology above can be drawn equivalently in many ways.  It can be drawn putting ANI just to the right of {Europeans}, to make ANI appear to be close to {Europeans}.  But ANI {32%-71%} is closer to {East Asian, Amerindian} than to {Europeans}.  That is the only way to read the 2016 paper, and that is how I have drawn the diagram, so as not to mislead.

4.  The last point is why don't people create charts like this?

What happened to the Europeans? 
Well, in the 2009 paper, what happened to the East Asians?  One can include Europeans as the dashed line below - notice that it is a very non-Eurocentric way of drawing exactly the same topology.


The answer is that the geneticists are, in my opinion, making some assumptions about history, largely because of the Indo-European language hypothesis, and perhaps because of the Colin Renfrew-ish hypothesis of a population explosion of the from-the-west-bringers-of-agriculture to India.

The 2009 and 2016 papers are not inconsistent.  But the 2009 paper exhibits the assumption geneticists make, not based on their own science, but from a particular reading of history.

PS: Since the commenters in my previous post simply don't get the point: suppose for arguments' sake, ANI was identical to European.  Then the 29%-68% ASI mix in the Indian groups would distance the Indian groups from European.   But it is surprising that the entire set of {European=ANI, ASI} mixes is closer to {East Asian, Amerindian} than to European.  
 
Based on historical biases, we should have expected Pathan, Balochi, Makrani to first form a group with Iranian, and so on. Sorry, they don't.  Even the Hazara, Shia and Persian-language speaking, are not close to the Iranian.   Likewise with the Tajik, who speak a Persian language (though the Tajik are largely Sunni).

Culture, language and genes are rather disjoint.






Comments (10)

Loading... Logging you in...
  • Logged in as
If you want to compare relationships, you need to look at shared nodes. The first node that Iranians share with Europeans is the same as the node they share with Indians (and all East Asians). So they are equally related to Europeans and Asians.
1 reply · active 441 weeks ago
And so?

Just FYI: half of Balochistan is in modern Iran.
The trouble with your final graph is not that it is non-Eurocentric. The problem is that it is wrong. The ANI are not more closely related to East Asians than they are to West Eurasians (whom you are calling Europeans). Just the opposite. The ASI (says the 2009 paper) are roughly as distant from the West Eurasians (and the ANI) as they are from the East Asians. The 2016 paper says that the current ANI/ASI mixes (as well as some tribal groups of partly East Asian ancestry) are closer to East Asians than to most West Eurasians. One plausible tree would have ancestral ASI as the founder population for all except West Eurasians, with East Eurasians splitting off from them very early.
3 replies · active 441 weeks ago
First part of my reply:

The 2009 paper calls the population "Europe (CEU)".
Diagram here: http://www.nature.com/nature/journal/v461/n7263/f...
CEU is : Utah residents with Northern and Western European ancestry (look it up!)

The 2009 paper is not talking about "Western Eurasians", because "Europe (CEU)" is not "West Eurasian" by any stretch of the imagination.
Hard to represent a tree here, but it would look something roughly like this, based on the 2016 paper:
{ {{ANI, ASI}, {East Asian, Amerindian}}, {Europeans, West Eurasians :) }}
The 2009 paper has (see the diagram http://www.nature.com/nature/journal/v461/n7263/f...

{ {{ANI, European}, {ASI, Andamanese}}, African}.

My last diagram still has, like the 2009 paper, the ANI splitting off the European branch.
The question is where does the East Asian group branch off?

I take the 2009 paper as saying that the ASI component of the Indian population causes it to be more distant from East Asians; therefore the group {ANI, ASI} can be close to East Asians only because ANI provides the closeness. That is, the subtree from the neighbor-joining algorithm { {ANI, ASI}, {East Asian}} is because of {ANI, East Asian}. That is why I have a branch of the tree splitting into Europeans and Asians, and then Asians splitting into East Asians and ANI.

I guess your comment is suggesting that the ASI component of the Indian population is what causes them to be close to East Asians. That is certainly worth thinking about, though I think the 2009 paper rules that out. The 2016 paper abstract also says: "These genomes include at least 5.8 million base pairs that are not present in the human reference genome", and that might be making some of the 2009 results a bit obsolete.
The last part of my reply: the 2009 paper relies on Principal Components Analysis, while the 2016 paper relies on neighbor-joining. I think we have to understand the peculiarities of these two different algorithms.

PS: perhaps one of the problems is that PCA reduces the dimensionality, and without a strict accounting of how much of the variation is accounted for in the first two principal components, we won't know whether a higher dimensional picture is what makes sense. The neighbor-joining algorithm doesn't seem to rely on a reduction of dimensionality.

PPS: The 2016 paper draws a 2 principal component diagram, the two components account for 7.8% and 4.0% of the variance. http://www.nature.com/nature/journal/vaop/ncurren...

The 2009 paper does a PCA on less diverse population. But the variance accounted for by the first two principal components is not mentioned anywhere as far as I can tell.
There are still a lot of puzzles concerning migrations of early humans, so the unbiased ought keep their options open. Consider the recent discoveries about Kostenki, and this quote:

Instead of inferring a few discrete migration events from Asia into Europe, we now see
evidence that humans in Western Eurasia formed a large meta-population with gene flow in
multiple directions occurring repeatedly and perhaps continuously.


From: https://www.repository.cam.ac.uk/bitstream/handle...
1 reply · active 441 weeks ago
I'm perfectly good with that. Note the "multiple directions" - this is a great improvement from the unidirectional flow into India that has been postulated till recently, with the Roma being a peculiar event.
I believe the problem lies with the tree model which are inadequate to represent human variation. A look at Indian Y & Mt DNA tells us that the most Indian Paternal lineages are Western Eurasian while the reverse is true with maternal lineages. Our plotting close to Siberian/N American Populations is on account of both groups possessing a combination of Eastern and non-Eastern Eurasian Ancestry and both groups being rich in ANE like ancestry. Have you seen the recent Lazaridis paper on Neolithic Iran? If New findings are to be believed we are essentially a combination of an ancient East Eurasian population (ASI), Neolithic Iranians and a Steppe population
1 reply · active 438 weeks ago
Right. Tree models do not correctly represent populations that emigrate and interbreed. We know a fair amount about population movements from Central Asia into Europe, but a lot less about movements into or out of India - mostly because DNA lasts longer in cold, dry climates than in hot, wet ones.

Post a new comment

Comments by