April 01, 2013

Directionality index for detecting origin of range expansions

This appears to be an interesting methodology for detecting directionality in genetic datasets. I am not sure how it might perform in the presence of admixture, a topic that was not discussed. Interestingly, the San appear as the only human population that had positive directionality values with all others, suggesting -to the authors- that they are closest to the origin of humans. On the other hand, in Pakistan, they found the Makrani to be the most ancestral population, and I strongly suspect that this may be related to the African admixture found in that population and not in others from that country.

arXiv:1303.7475v1 [q-bio.PE]

Detecting range expansions from genetic data

Benjamin M Peter, Montgomery Slatkin

We propose a method that uses genetic data to test for the occurrence of a recent range expansion and to infer the location of the origin of the expansion. We introduce a statistic for pairs of populations $\psi$ (the directionality index) that detects asymmetries in the two-dimensional allele frequency spectrum caused by the series of founder events that happen during an expansion. Such asymmetry arises because low frequency alleles tend to be lost during founder events, thus creating clines in the frequencies of surviving low-frequency alleles. Using simulations, we further show that $\psi$ is more powerful for detecting range expansions than both $F_{ST}$ and clines in heterozygosity. We illustrate the utility of $\psi$ by applying it to a data set from modern humans and show how we can include more complicated scenarios such as multiple expansion origins or barriers to migration in the model.

Link

10 comments:

eurologist said...

Seems like a pretty straightforward approach, but the devil as always lies in the details, and their Fig. 9 clearly indicates this is not ready for prime time, yet. I don't think the authors will find much support that S Europe was settled directly from the southern Levant, and that most of W Asia and S Asia (and SE Asia, indirectly) was settled from Europe...

Until relatively recent admixtures can easily be filtered out, a more reasonable approach would be to be more selective in the populations (robustness can always be tested by including/excluding others) to avoid known admixture issues.

For example, use as a subset E African, European, Pakistani, Indian, and E/SE Asian population. I bet the directionality to and from Europe would reverse.

Comrade Dos said...

Dienekes -- I would like to interview you for a news article about the use of tools like Dodecad to reassess microarray data from ancestry testing services. I work for a New York-based media company called GenomeWeb. Please contact me at jpetrone@genomeweb.com to follow up.

DMXX said...

Yes, it possible that the Sub-Saharan African ancestry of Pakistan's Makrani has yielded a false result with regard to their range expansion rating. A simple but powerful deduction here.

However, as per the Out of Africa (OoA) model, the Makrani coast lies directly on the path that was presumably taken by one of the two primary waves of anatomically modern humans (AMH's).

It is therefore possible the researchers have stumbled onto a genetic sign of diversity that autosomal component software (i.e. ADMIXTURE) is not built to detect.

eurologist said...

"However, as per the Out of Africa (OoA) model, the Makrani coast lies directly on the path that was presumably taken by one of the two primary waves of anatomically modern humans (AMH's)."

DMXX,

That coast and surroundings of Balochistan were uninhabitable during very long stretches between 70,000 and ~15,000 ya. It is highly unlikely that anyone from the initial ooA movement(s) survived there or in the vicinity.

terryt said...

"Another interesting result is that the Melanesian and Papuan samples, while very similar, show positive values when compared to
other East Asian populations, but the directionality index is negative when compared to the Pakistani, European and African populations. This is consistent with a 'two-wave' model of colonization of South-East Asia, with a first wave consisting of present-day Papuans and Melanesians, and a second wave
consisting of the present day Chinese populations [53]".

Or possibly indicates two separate routes of migration. One south of the Himalayas and the other to their north.

"use as a subset E African, European, Pakistani, Indian, and E/SE Asian population. I bet the directionality to and from Europe would reverse".

I would like to see West African and Australian examples included as well. And East Asian, centred on the region of the EDAR370A origin, included rather than the obviously admixed SE Asians.

terryt said...

"That coast and surroundings of Balochistan were uninhabitable during very long stretches between 70,000 and ~15,000 ya. It is highly unlikely that anyone from the initial ooA movement(s) survived there or in the vicinity".

In fact I think it highly unlikely that any initial ooA movement(s) used that route at all.

eurologist said...

In fact I think it highly unlikely that any initial ooA movement(s) used that route at all.

Terry,

There were numerous time intervals between ~130,000 and 70,000 ya when for tens to hundreds of years, the coast line and surrounding hills had sufficient wadis, wells, grass land, and associated animals that it would have been an easy path for AMHs with limited skills and adaptations to follow. But not immediately before, and not after - until very recently.

terryt said...

"There were numerous time intervals between ~130,000 and 70,000 ya when for tens to hundreds of years, the coast line and surrounding hills had sufficient wadis, wells, grass land, and associated animals that it would have been an easy path for AMHs with limited skills and adaptations to follow".

Yes, but ... at such times the inland regions would be even more inviting. It is surely far more likely that at such times human groups spread right through much of Arabia and Iran. They would have had no need to hug the coastline.

eurologist said...

"...the inland regions would be even more inviting. It is surely far more likely that at such times human groups spread right through much of Arabia and Iran. They would have had no need to hug the coastline. "

No - the Iranian plateau surely was the worst place to live in even in the few good times, as was southern Afghanistan. In addition, records show Neanderthal or Heidelbergensis-like people there, whenever inhabited, until about 50,000 years ago.

terryt said...

"records show Neanderthal or Heidelbergensis-like people there, whenever inhabited, until about 50,000 years ago".

Both types certainly occupied regions far to the north of Iran, let along Makran. Although how they reached that region reminas unknown as far as I'm aware.

"the Iranian plateau surely was the worst place to live in even in the few good times, as was southern Afghanistan".

I find that hard to believe. In fact I have been unable to find any reference as to the Makran coast being more suitable as human habitat than are regions further inland at any time. Do you have information on the subject? Certainly people occupied Afghanistan periodically, as this book shows:

http://unesdoc.unesco.org/images/0009/000944/094466e.pdf

In the section 'Hindu Kush: the mountain region between the Indus
and the Oxus', quote:

"Likewise, Bisitun in the Zagros mountains; Shanidar excavated by
Soleki in 1971, with a Mousterian industry and a spectacular Neanderthal burial accompanied
by red ochre, ibex horns and flowers; the many other sites of this region and of the Crimea are all part of the major geographical and environmental regions. They do however form part of a continuum of associated Mousterian industries and Neanderthal physical
remains that stretch from T’ien Shan through central and western Asia, southern Russia, eastern and western Europe and north Africa to the Atlantic".

That places the east/west connection far to the north of Makran. And here:

http://en.wikibooks.org/wiki/Iranian_History/Ancient_pre-historic_Iran

Quote:

"The earliest known evidence of hominin presence in Iran consist of a number of occurrences with surface collection of lithic tools that fall within Oldowan and Acheulian stone tool traditions. These sites are Kashafrud (Khorasan), Ladiz (Sistan and Baluchistan), Ganj Par and Darband Cave (Gilan), Shiwatoo (Mahabad), and few other sites in western Iran such as Amar Merdeg, Pal Barik and Gakia".

Ladiz is hardly 'coastal'.