December 29, 2011

North African genetics through the prism of ADMIXTURE

I believe that with this exercise, which took me just a morning's time, I'm walking a path that has not been explored before: analyzing the autosomal genetics of North Africans on their own right, without  being part of a larger context, be it African or West Eurasian or global. At least I'm not aware of any such paper nor self-research exercise in the blogosphere either.

Said that, I did get in the study five exogenous samples, in order to estimate possible external influences. These are: 10 Fulani, 10 Mandinka, 10 Ethiopians, 10 Saudi Arabs and 10 Spaniards. I did not alter the diverse HGDP North African samples (including two different Egyptian samples), except for two things: I removed the Moroccan Jews altogether and cut the Mozabite sample to 10 individuals, because of suspicion that their alleged isolation might distort the larger analysis.

More or less as I expected, at K=10, which was my preliminary goal, each of the exogenous ethnicities described one distinct component, while the other five components were North African specific.

What I did not expect at all was that Tunisians would show up as distinctive as they did (see below). I wonder if there is something special in that sample or if the measure applies to all (or most) Tunisians. Very strange and unexpected in any case.

In the end, concerned that I might be missing something of relevance, I made two more runs and one of them struck "genetic gold", it seems to me: a small South Moroccan component very distant from everything else, which might well be a remnant of the Aterian period or something like that.

Method: I used a fraction (as described in the previous lines) of the global HGDP sample following the method explained at Gene Expression to operate ADMIXTURE and associated programs (Plink, R). 

Results:


K=2 - Without surprises: Tropical African vs. West Eurasian components.

K=3 - Big surprise: the first North African specific component is concentrated in Tunisia, not Morocco, not Mozabites... but Tunisians, uh?

K=4 - As another North African specific component (red, most common among Sahrawis, then Moroccans) shows up, the Tunisian component (green) retracts, so to say, to the Tunisian borders.

K=5 - Not happy with one, the algorithm finds a second Tunisian component, restricted also to that country. I'm as perplex as you may be.

K=6 - West Asian (Saudi Arab) and European (Spaniard) components diverge.



K=7 - Second (non-Tunisian) NW African component shows up. This one (turquoise) is most concentrated among Mozabites. 

K=8 - A Fulani-specific component shows up. Intriguingly it is almost equidistant by Fst measure from the Mandenka and the Sahrawi components (0.105 and 0.115 respectively). All the North African specific components are much closer to West Eurasian ones than to the Mandenka component, so this might suggest a very old kind of trans-Saharan admixture, then homogenized in a single component. 

K=9 - Not happy with one, the Fulani show a second component in a row. This one is neatly Tropical African (very distant from all and only somewhat close to the Mandinka component and the other Fulani component but at the 0.163 and 0.173 Fst values, which is also very distant). I imagine that this has to do with the Fulani L1b mtDNA lineage but never mind because the component will vanish again as we move on. 

K=10 - A Morocco-centered component shows up here (green), also found in Algerians and Libyans. A distinct Ethiopian-specific component is also defined (influencing Egypt and Libya significantly and to much lesser extent also NW Africa).

K=11 - A small and very interesting component exclusive of South Morocco shows up. 

(Note: at K=12 there is a third Tunisian component, go figure!, but I don't think that is informative at all so it's not shown). 

Note: A reader suggested that some North Africans in these samples are heavily admixed with Tropical Africans, distorting the results in that aspect. I can't say but, if I manage to get working the program variant that should show individual instead of population bars, then we will find out.

Fst Distances at K=11:


Notice please that the South Moroccan component is extremely distant to all (Eurasians and Africans alike). I will speculate (as I have done before seeing this) that this component, now almost only restricted to Southern Morocco and heavily admixed, is a residue of the Aterian period and is related to a vaguely "Khoisanid" or equally vaguely "Mongoloid" phenotype found in the region.

Component apportions (numerical) at K=11:



Detail of K=11 graph:


47 comments:

  1. i think the tunisian sample is inbred. i've seen the same. how many markers did you have left?

    ReplyDelete
  2. I also think that the Tunisian sample is strongly inbred. Otherwise it does not make much sense and Tunisians (same sample?) have already appeared as very non-diverse in previous studies. I was very bothered by this phenomenon but I can't do much about it.

    "how many markers did you have left?"

    No idea what are you talking about. How can you figure that out?

    ReplyDelete
  3. http://forwhattheywereweare.blogspot.com/2011/03/african-diversity-and-possible-origins.html

    Scroll down and you can see how Tunisians strike as highly non-diverse. I wonder if it is the same sample or it is just that all (or most) Tunisians are that way. Shouldn't be: I'd rather imagine Tunisians as the most cosmopolitans of all North Africans, so it's probably that we are dealing with a bad sample.

    ReplyDelete
  4. "Notice please that the South Moroccan component is extremely distant to all (Eurasians and Africans alike). I will speculate (as I have done before seeing this) that this component, now almost only restricted to Southern Morocco and heavily admixed, is a residue of the Aterian period and is related to a vaguely "Khoisanid" or equally vaguely "Mongoloid" phenotype found in the region. "


    It is as distant as the odd Slovene component! I think that's a good explanation. What does suprise me is that Spaniards don't seem to have any N. African admixture except for the S. Moroccan, which is minimal (0.2%), that doesn't agree with other analyses in which most Iberians show varying % (from 5 to 10%, except in Basques) of N. African admixture.

    ReplyDelete
  5. This sample specifically does not seem to have significant North African admixture (in the previous run it did only show 2.3% of the North African component) however now it looks even less admixed - I guess it may be caused by none of the various components being too relevant (I would not give it much importance: in any case North African admixture in Iberia is low but real). I imagine that a Portuguese or Leonese sample would but I also expect a West to East cline in that, so Eastern Iberians (2/3 of the peninsula rather than just the Eastern half) should have very low North African admixture.

    We were just discussing this matter at Razib's blog and there seems to be a compatibility problem (different SNPs tested for) with the IBS and the HGDP samples, so it won't be easy to find out how the North African influence spreads through Iberia. I suggested using Canarians as proxy for North Africans (they are more Iberian than North African, except maybe the Gomeros, but they still retain more Berber blood than any true Iberian for sure).

    ReplyDelete
  6. My impression is that North Africa should be thought about as a series of islands, isolated during most of the late glacial periods due to the extremely arid climate (lack of surface water). After the Aterian, the early upper Paleolithic Dabban (42-30k BP) is only represented in Lybia. Most of North Africa was probably uninhabited after that. Archaeology sites indicates the next group of people (outside of Egypt were the Iberomauserian (15-10k BP) who moved in after the last glacial maximum. I wouldn’t expect any genetic influence from the Aterian to be remaining in North Africa.

    The biggest population jump was probably Caspian Culture and during the Neolithic Pluvial (8000-4000 BP) when North Africa better connected and the Sahara contained large lakes and river systems. The dates of the various industries are variable.

    There were two different types of people according to the skeletal evidence, a "robust" group descended from the Iberomauserians and a "gracile" group, more recently from the Middle East. I would expect these groups to have different genetics, but I don't know of any DNA data from them.

    ReplyDelete
  7. David (Eurogenes) keep only five Mozabites, the most outbred and non-SSA admixed, and it worked well (no Mozabites-centered component) at K=9:
    http://eurogenes.blogspot.com/2011/11/on-origins-and-expansions-of-r1a-and.html

    Tunisians are of course inbred too. And the 'southern Moroccan' group is very heterogenous and it implies a lot of SSA admixed and even some mostly-SSA individuals. The 'Moroccan' group has 3 outliers.
    Sahrawis too have SSA admixed individuals but strikingly less than southern Moroccans, despite their location (closer to west africa).

    Anyway, we can ignore the recent ssa element by focusing only on Northern Moroccans, Algerians and Tunisians but the latter is too inbred.

    ReplyDelete
  8. @Argent:

    I do not think that the aridity should be so extreme and I rather think of the region as a single mountainous/coastal area, separated from the rest of the World by deserts (except in pluvial periods) and the sea. In the Ice Age the climate should be much like that of Mediterranean Europe, i.e. thick forest cover. The fact that the region has hosted specific fauna and flora related to that of Africa or Eurasia but also distinct (Berber macaque, Atlas cedar, elephants, etc.) strongly suggest that it was habitable all the time. Not the Sahara (except in pluvial periods) but certainly most of the area that is densely populated today between the Atlas and the coast, which is of Mediterranean climate.

    Dabban industries are only known in Libya, yes, but that may be because of lack of sufficient research. Genetically North Africans (and very specially Moroccans) do appear to have some marked "West Asian" influences from that period, like mtDNA U6 (whose basal diversity is highest in Morocco and, only after Morocco, in Canary Islands and Iberia). I don't think this first West Eurasian layer in North Africa should be attributed to Iberian flow, not yet.

    Oranian has its oldest dates in Morocco, so the claim that it is of Egyptian origin seems to me a construct without substance. Also there are obvious similitudes beytween Oranian the original Gravetto-Solutrean (or Solutrean of Gravettian substrate and influence) of Iberia and we can even detect the first known usage of tanged and winged points in Iberia in the late Solutrean period, possibly because of back-influence from North Africa, where it is a tradition beginning with Aterian.

    Also there is a lot of Iberian-derived genetics in North Africa which need an explanation. Mine is that they arrived with the Solutrean-Oranian interaction, which is the very genesis of Oranian culture (once known as Iberomaurusian with some good reason).

    "Caspian Culture".

    Capsian! (Capsa = Gafsa, Latin version).

    This one did indeed originate in Nubia (border area between Egypt and Sudan). Sadly I did not have a Nubian sample so I used Ethiopia as proxy (not a very good one, admittedly). There is no clearly distinct Neolithic from Capsian and the concept of "Capsian Neolithic" (i.e. a second phase based on Capsian Epipaleolithic that incorporates farming and herding) is common nowadays.

    "There were two different types of people according to the skeletal evidence, a "robust" group descended from the Iberomauserians and a "gracile" group, more recently from the Middle East".

    The same idea of a native "robust Mediterranean" and a colonist "gracile Mediterranean" has been used in Iberia but I wonder how solid it is, specially as diet alone can alter the height and other skeletal elements.

    ReplyDelete
  9. @Idurar: Do you think that the Mozabite sample is bothering the analysis? I was much more upset by the results regarding the Tunisian sample. The Mozabites are an isolated community and will eventually get their own distinct component (it's unavoidable).

    I guess it may be of some interest to re-run the sample without Mozabites and Tunisians but I'll leave that for a later occasion if anything. This is a first approach.

    Honestly I'm more concerned by the fact that such a large (double) sample as the Egyptian one shows no distinct personality at all, appearing as a tripartite mixture of Arabs, Iberians and Ethiopians... how come?!

    Probably there is even more to dig and refine in North African genetics but I also feel that as first approach this one is better than nothing (I have seen no such analysis before, have you?)

    "The 'Moroccan' group has 3 outliers".

    Maybe. I could not yet get to work the individual bar script that Razib sent to me so I'll have to trust your word. But there are two other Moroccan samples (North and South). As anyone could expect the sample showing the greatest trans-Saharan admixture is the Southern one, and the one showing the least is the Northern one, with the "Moroccan" (Central?) one being intermediate. Inversely for the Iberian admixture, also in good logic.

    "Sahrawis too have SSA admixed individuals but strikingly less than southern Moroccans, despite their location (closer to west africa)".

    Sahrawis have that trait and also a distinct component of themselves, which is somewhat close to the Fulani one but also within the West Eurasian group (that's a very curious "inter-racial" transition that the Fulani component makes). Otherwise they seem more isolated than Southern Moroccans, what may be related with Marrakesh being an important city and trade center along the centuries, something that the Sahara could never be. Also the main trans-Saharan route by the West seems to have gone through farther East than the West Sahara, through the Taghaza salt mines, in Tuareg territory.

    ReplyDelete
  10. Tunisia
    i was going to say Carthage but you'd expect some sign of that elsewhere also.

    "The same idea of a native "robust Mediterranean" and a colonist "gracile Mediterranean" has been used in Iberia but I wonder how solid it is, specially as diet alone can alter the height and other skeletal elements."

    Doesn't diet make sense as the reason for it?

    Islands of coastal/river farming settlements with high pop. density but small people amongst a sea of
    bigger herders / foragers / fishers with a higher protein diet but lower population density.


    "such a large (double) sample as the Egyptian one shows no distinct personality at all, appearing as a tripartite mixture of Arabs, Iberians and Ethiopians... how come?!"

    Perhaps once you've reached a very large, high density farming population it's simply very hard to impact it significantly.

    ReplyDelete
  11. ''http://forwhattheywereweare.blogspot.com/2011/03/african-diversity-and-possible-origins.html

    Scroll down and you can see how Tunisians strike as highly non-diverse. I wonder if it is the same sample or it is just that all (or most) Tunisians are that way. Shouldn't be: I'd rather imagine Tunisians as the most cosmopolitans of all North Africans, so it's probably that we are dealing with a bad sample.
    ''

    These Tunisian samples are isolated Berbers, not cosmopolitan Arabic speaking Tunisian.

    ReplyDelete
  12. The Mozabite and Tunisian Berber samples always tend to cluster away from the main mass of Maghrebi.

    ReplyDelete
  13. It is a nice exercise but is *very* difficult to follow these graphs.

    ReplyDelete
  14. @Antonio Pedro:

    "... is *very* difficult to follow these graphs".

    True. I blame that on two reasons:

    1. That neighbor or akin samples are not grouped together. I don't know how to work that out but it should be possible. Bear with my clumsiness or help me solve these issues.

    2. Colors: as it is the program choses a palette that is always red for Pop0 and equally spaced hues for the rest. While this is technically correct from a perspective of "real" colors, the human eye is not equally good discerning all shades and is actually quite bad with the greens and blues, which tend to look all the same. Never mind also that most of these colors are less ugly and unnatural in darker shades.

    I wish I could change this also but, if I re-paint the components I risk committing human errors, what would be worse. Also dyeing the bars and re-sorting the populations manually would take more time than I'm willing to dedicate.

    So if there is someone with some basic programming knowledge that can help me to sort all that out with one or few scripts, I'd very much appreciate it.

    ReplyDelete
  15. @Awi:

    "These Tunisian samples are isolated Berbers, not cosmopolitan Arabic speaking Tunisian".

    I see. They are a very special population: very isolated or "inbred", it seems.

    I may try in the future an analysis without the outliers anyhow.

    "The Mozabite and Tunisian Berber samples always tend to cluster away from the main mass of Maghrebi".

    Are there other similar studies on North African population structure. You speak as if there were many yet I am unaware of them: they are always placed within larger continental or global analysis, what is exactly what I meant to overcome with this exercise.

    ReplyDelete
  16. @Grey:

    "Perhaps once you've reached a very large, high density farming population it's simply very hard to impact it significantly".

    That makes sense on its own right but I fail to see how it relates to the apparent lack of "personality", distinctiveness, of the Egyptian sample. Shouldn't those farmers have coalesced at least locally into more or less homogeneous components since Neolithic? Why is there such a huge European ("Iberian") influence (probably a mirage)? I am very perplex at the two Egyptian samples (which perform almost identically).

    ReplyDelete
  17. Hello Maju,
    I found the same thing with the Tunisians when I did a K=10 intra continental African run a while back:

    Selected Average

    All Average


    Can't really say what was up with them though.

    ReplyDelete
  18. Thanks for the info.

    We know per Henn 2011 (which is the origin of this sample) that they have less diversity than the Hadza, which also produce a similar phenomenon at East African level (they quickly coalesce into their own, cluster), and which are at the diversity level of other North Africans (makes sense because North Africans are essentially Eurasian genetically speaking and Eurasians have much less genetic diversity than Africans).

    So whatever is wrong with the sample, it's clear that its main problem is low diversity, what creates a mirage of uniqueness - technically correct but useless in practical terms, specially when they begin cluttering K-levels one after the other.

    ReplyDelete
  19. maybe true about the limited genetic diversity of the Tunisian samples skewing the results, I would however be less inclined describing North Africans as 'Essentially Eurasians', since they are rather genetically intermediate between other Africans and Eurasians, if global principal component analysis is any indication.
    Also, I would recommend that you run a PCA plot on each data-set that you are running ADMIXTURE on so that you will have a more complete picture of what the data says.

    ReplyDelete
  20. "I would recommend that you run a PCA plot"

    I don't know how to do that. Any suggestion?

    In any case I find PC analysis often misleading because they hide nearly everything that is not described by the two first components, so it's like looking at a K=3, more or less, with all its undesirable oversimplification.

    "I would however be less inclined describing North Africans as 'Essentially Eurasians', since they are rather genetically intermediate between other Africans and Eurasians, if global principal component analysis is any indication".

    I do not think that idea is correct. Even if North Africans do have some tendency towards Tropical Africa, they clearly cluster much more closely with West Eurasians.

    In this case, lacking the PCA, we can look at the Fst distances: excepting the equally distant (and most interesting therefore) South Morocco component, the distances with the Iberian and Arab components are always much smaller than with the Tropical African ones.

    Distances to external components: Iberia, Arabia, Ethiopia, Fulani and Mandenka (in this order and in milesimal figures, i.e. 89 is actually 0.089):

    Tun1: 89, 103, 101, 152, 182
    Sah: 100, 107, 104, 156, 178
    Tun2: 120, 157, 130, 189, 214
    Moz: 116, 128, 120, 174, 197
    Mor: 74, 81, 71, 122, 148
    (Average: 100, 115, 105, 159, 184)

    So the local NW African components are all more akin to (1) Iberian, (2) Ethiopian (one is exceptionally inverse for this two only), (3) Arab, (4) Fulani and (5) Mandenka. And the difference with the last two is of the order of 50-85%, quite notable.

    It is true that the Ethiopian component is intermediate between the two continental areas (this is truly half-way in fact) and I have to admit that there are no other East African population to compare. But judging from the Mandenka and Fulani alone, NW Africans are much closer to West Eurasians.

    The Ethiopian component and its Tropical Africanness (at least 1/3 judging by K=9 but Mandenkas are a poor proxy surely) can't be ignored however but this is a different kind of "Africanness" than the usually inferred from comparison with West Africans. The proximity to Ethiopians in Fst probably represents the Capsian influences, while the proximity to Iberians surely reflect Oranian ones.

    In addition, North African populations have some 15% of strictly West Eurasian and also some 10% of strictly West African blood, judging from this analysis at K=11.

    And then there is the S. Moroccan component which is a wholly distinct continental population in their own right, IMO.

    But well, I'm just exploring multiple avenues, K=2 should have the simplest answer to this B&W (literally!) question. And the answer is that North Africans have some but rather low affinity with the Tropical African component polarized in the Mandinka, maybe 15% on average.

    But even Ethiopians score low in this: just c. 35%, so maybe it's not totally fair, maybe we're comparing apples (low diversity East-African-derived West Eurasians) with oranges (high diversity Tropical Africans).

    It is an interesting question but I still think that North Africans cluster much more closely with West Eurasians than most other Africans.

    ReplyDelete
  21. It is pretty easy to do a PCA or MDS, they are both pretty much identical, except in MDS you have to specify the number of dimensions, whereas a true PCA would yield you N dimensions from an N X N matrix. Anyway plink has an MDS command and that is what I used in the past, refer to this link for additional details :PLINK MDS

    Here is a summary of the steps you need to take:

    First make a “genome” file. Assuming the .bed file that you want to do an MDS on is called “African_plus_me”, and the genome file you want to output is called “Africa_Distance_Based” then type the following command:

    plink --bfile "African_plus_me" --genome --out "Africa_Distance_Based"

    You will now have a genome file called “Africa_Distance_Based” in your folder, then type:

    plink --bfile "African_plus_me" --read-genome "Africa_Distance_Based.genome" --cluster --mds-plot 4

    The last number you see in bold in the above command is the number of dimensions you'd like to generate, if you just want 3 then change it to 3, etc. So the above command will generate you a file called plink.mds. Basically the mds coordinates for each individual in your data-set, then you simply plot the coordinates, it is your choice what graphing software you use, I use GNU octave (the open source version of matlab) to plot my results, 3D plots are more illuminating than 2D plots by the way.

    Now let me briefly try to get to some of your other points,

    1) “In any case I find PC analysis often misleading because they hide nearly everything that is not described by the two first components...”
    PCA analysis is important because it is distance based rather than model based (ADMIXTURE for instance) they both complement each other and should be done together. If you notice in most professional papers they present both a PCA plot alongside the STRUCTURE (or whatever program they use) analysis. I agree that missing components might hide something, but that is upto you to make sure that you look at all the dimensions you think are important, also the dimensions becomes less and less important in describing variation as they increase in number, thus, the first dimension is more important than the second, second more important than the third and so forth, that is how the mathematics of ordination works.

    2) “Even if North Africans do have some tendency towards Tropical Africa, they clearly cluster much more closely with West Eurasians.”
    But I never said anything about Tropical Africa, I said other Africans, other than the North Africans being discussed that is, what exactly do you mean by Tropical Africa? places on the continent between the Tropics of Cancer and Capricorn ? If so that is a huge landmass, with a vast variety of people, it seems as though you are alluding to forest Africans maybe? Equatorial Africans ? Humid Tropics? Sub humid tropics? With out a precise definition of what you mean by Tropical Africans it is rather hard to understand what you are talking about.

    ReplyDelete
  22. "Why is there such a huge European ("Iberian") influence (probably a mirage)?"

    My assumption is if the first farmers started out in Asia Minor then the same component that went west to Greece and North to the Caucasus also went south to Egypt. If so then the Iberian component is being picked out as the closest match because the Iberian component also has some of that first farmer element but the Egyptian sample would join a Greek or Caucasus component first if either of them was in the list of samples.

    If so the Egyptian compound would be an early layer of those guys overlaid with an African element from the south and an Arabian element from the east.

    "Shouldn't those farmers have coalesced at least locally into more or less homogeneous components since Neolithic?"

    That's the intriguing thing. I had the expectation that these major genetic elements would kind of bleed together over centuries whereas currently it looks more like chemistry where you have elements like Carbon and Oxygen and they can combine in many ways e.g. C2O or CO2, but the elements themselves somehow remain separate.

    Maybe that's just a function of the software?

    ReplyDelete
  23. @Etyopis:

    Thanks for those most comprehensive and detailed instructions. I will try to follow them even if just for getting familiar with the process.

    However I still think that PC graphs, even 3D ones, miss most of the substance and are only apt for over-simplified, rather shallow, analysis.

    They are easier to read, so nicely visual!, admittedly, but they compare to structure-like analysis very badly.

    "but that is upto you to make sure that you look at all the dimensions you think are important"...

    But that is even more confusing because you do not really know what is real and what an artifact when you look at dim 7 vs dim 3 for example. As more dimensions are extracted also, the number of possible representative graphs grows exponentially.

    I mean: it's a tool but not my favorite at all. What I would like is to get an individual bar plot instead of these whole population ones, that may be very visually informative in some cases.

    "... also the dimensions becomes less and less important in describing variation as they increase in number, thus, the first dimension is more important than the second"...

    Indeed but it depends on what they actually represent, how much do they weight in each case. For example PC1 may be 16% or may be 2%, what means totally different things. PC2 is always smaller but it can be 15% (very important) or 0.2% (irrelevant noise).

    This should affect how graphs are stretched and IMO both axes should never be represented of equal length (unless they are, what is very rare), because it induces the wrong impression: that all plot distances are equal genetic distances, when they are often very different.

    I discussed this matter years ago: http://leherensuge.blogspot.com/2008/02/something-to-ponder-genetics.html

    In that case we had an African-Other PC1 axis representing more than 50% of the detected variation and a West-East Eurasian axis representing barely 25%, yet they were both the same size, so I applied a quite radical (but accurate and necessary) correction to the representation for the sake of learning something.

    I also mentioned a bit later Novembre and Stephens 2008, who argue that PC analysis is limited in inference power.

    ... (continues)

    ReplyDelete
  24. ...

    "But I never said anything about Tropical Africa, I said other Africans, other than the North Africans being discussed that is, what exactly do you mean by Tropical Africa?"

    Other Africans than North Africans. I strongly dislike the term "sub-Saharan" because I find it eurocentric and racist and geographically incorrect and worse: people are swallowing it as if it was "politically correct". "Sub" means "under" (as in sub-human, submissive, submarine, etc.) and "under the Sahara" there may be rock, magma, oil, water... but excepting for some old corpses there is no people.

    Trans-saharan would still be eurocentric but at least geographically correct (nope: the south is not below the north, Madrid is not "under" Bilbao, if anything it's the other way around, as Madrid is 500 m. above sea level). The mental deformation caused by North-on-top Mercator projections is scary!

    I would use the classical Black Africa but that is also imprecise and some people would consider it racist, so I tend to use Tropical Africa (when needed adding "and Southern Africa").

    The Tropics are the area between the Tropic of Cancer (Sahara) and Tropic of Capricorn (Kalahari), so it gets most of that "other Africa". "Tropical Africa" (or "Tropical and Southern Africa" when necessary) is the term I prefer to use to that horrible "subsaharan" nonsense.

    See also: http://leherensuge.blogspot.com/2008/05/supersaharan-africa.html (admittedly a ranting but a cool one anyhow - I always think I want to make a T-shirt out of that map and then I forget).

    "places on the continent between the Tropics of Cancer and Capricorn ? If so that is a huge landmass, with a vast variety of people"

    Yes, I mean that. And I agree that it is huge and that it has a vast diversity of people.

    It is not less precise than when you say "other Africans" anyhow.

    So we can go precise whenever you want (provided that a manageable sample is available and we have an idea that we know what we want to find out), I just love to be precise and not using sloppy terms, but if you speak of "other Africans" I can well speak of "tropical Africans" all the same (and I'm at least excluding Egyptians and San, not much of a narrowing but something).

    Now if you propose that NW Africans are more similar to the Sandawe or the Luhya or some Cameroonian group... we can study that with some good precision. But I think in advance that you would be wrong. A very different case could be if we could have some Nubian or Sudanese sample; there we might find more similitudes, but that is a very specific area and not the vague "other Africans".

    ReplyDelete
  25. @Grey: you are assuming migrations of farmers to Egypt. I do not think that there is any archaeological data of any sort that supports any such thing: no PPNA nor PPNB in Egypt... instead there may have been Egyptian influence in the Mesolithic of Palestine (Kebaran, Harifian and such).

    However it's possible that the Iberian component is just closer to, say, the Caucasus (Highland West Asia) component than the Arab one...

    ... or rather that the generic West Eurasian component present is best expressed as a 50-50 mix of Iberian and Arab, while being akin in fact to neither of them.

    ... "but the elements themselves somehow remain separate".

    Not always. There is the possibility that Admixture is just not getting the mixed compound at the levels and sample numbers used. We have an example in the Fulani: a first they look a mix, then they display an equidistant component of their own (the equidistance also suggests ancient "homogenized" admixture of the kind you expected).

    After all, the program just indicates "estimated affinity" and has to discern distinct components one by one: it's not rocket science.

    ReplyDelete
  26. @Maju
    "you are assuming migrations of farmers to Egypt. I do not think that there is any archaeological data of any sort that supports any such thing"

    I tend to look at it backwards. Logically, first farming in Egypt had to either be developed independently or arrive from North, East, South or West so i assume each possibility in turn and then see if any of them can be eliminated.

    I add the assumption that as farmers can't move and grow food at the same time they'll move in ways which don't take too much time i.e. either very short distance diffusion from an existing settlement, or longer distance but fast hops by water to create a new settlement.

    So for me that gives the possibilities:
    1) indigenous
    2) up the Nile from the South
    3) med. coast from either north, west or east then down the nile from the north

    http://en.wikipedia.org/wiki/Prehistoric_Egypt

    In this case Wiki says almost all neolithic finds in Egypt come from the south however it also says anything in the north is liable to be buried under the delta silt so that doesn't help.

    ReplyDelete
  27. "Logically, first farming in Egypt had to either be developed independently or arrive from"...

    Sure but for the same logic, did the arrival of industrialization to Russia or Japan imply a demographic replacement? Not at all: the knowledge spread with human interaction but not necessarily and not in most cases demographic colonization. So that argument is a just a fallacious one, allow me to say.

    Now, it's not impossible but I'd like to know with more certainty.

    ReplyDelete
  28. Are you using R for the graphs? For one thing you are trying to accomplish too much at once. You cannot plot one graph on the top of the other, each one for a different k, without making the colors consistent. If the colors cannot be consistent, then choose different colors for different k's. If not possible, then have some space between the graphs. You can go up in terms either of the number of individuals in these plots or the number of ancestral components. But one cannot pair several plots with inconsistent color scheme. But you don't have to do that anyway. If you cannot come up with something new, which is fine, just do what people is doing in terms of graphing these results.

    I don't think you have a programming problem, at least not now. Once, you have a clearly idea of what you want to graph then you might have a programming problem. But that's not the case yet.

    ReplyDelete
  29. Yes, Antonio, I'm using R but rather blindly at the moment. i have managed to understand Razib's instructions and use his "source" script but that's about it so far. I haven't even read the user's manual nor I'm even sure it has one.

    Totally short-term goal oriented, admittedly.

    "You cannot plot one graph on the top of the other, each one for a different k, without making the colors consistent".

    So there is a way to make the colors consistent: that sounds nice. Now the problem is how to do it because each thing that R does seems to require a script and my knowledge of programming is as rusty and useless as Cobol, the language that would never link (almost no idea of Javascript or C++ or whatever its used with R - I can't even discern one from the other, really).

    "But one cannot pair several plots with inconsistent color scheme".

    You are logically right: it's aesthetically horrible and conceptually hard... but HOW TO?

    "... just do what people is doing in terms of graphing these results".

    If I can do half of half, what is what I am doing now, I'm happy. I would like to do better but I am not qualified right now.

    So sorry about the colors (again) but that's what I know HOW TO do.

    If you want something else, teach me HOW.

    "Once, you have a clearly idea of what you want to graph then you might have a programming problem. But that's not the case yet".

    I think I have got a programming problem because I know I want some things like neat colors, individual bar plots and so on but I can't do that yet. So it's a script or programming inability problem what is causing those ugly colors (and other lacks).

    ReplyDelete
  30. I am not criticizing you. On the contrary, I applaude your effort. My point is very simple: you're indeed trying to do more with bar plots than other bloggers do. Consider, for instances, Dodecad or Harappadna, which make extensive use of this barplots in a elegant way. They never stacks one over the other. The thing is that when you change the k's you force the analysis to produce different ascentrals components for the "same" population. Thus, when you have k+1 the meaning of the previous k changed. Thus even if you can keep the same colors from the previous analysis and just addd one more, that might still be misleading for in the new analysis, say, the red color will denote a different ancestral component. Thus we have a conceptual problem, not a programming problem. You might want show how population diverges as the number of ancestral components change. One way to do that, is just have separate barplots, discussing one at the time, as the other "serious" projects do. That's my suggestion for you now. Another way is to create a different graph, focusing on the dynamics of change per sei. This will be more sophisticated solution. If you wanna do that, first you have an conceptual problem. Then, after figuring it out the new plot, you will have a programming problem, which I might be able to help. Cheers, Antonio.

    ReplyDelete
  31. We do not have a conceptual problem, stacking K barplots have been used since "always" in the literature, just that they manage to get a more consistent color system, what I haven't managed to achieve yet. I have more the literature as reference than Dodecad of Harappa, which tend to do global analysis instead of regional, use "zombies" an so on. I look to everywhere I can learn from but my real problem (as per your criticism, which is essentially correct) is not stacking barplots (what IMO is a must for the reader to be able to compare different K levels and not just be directed to think this or that by a carefully selected layer of analysis) but not to be able to use a more beautiful and consisten color system. I could try dyeing them by hand but I'm normally too anxious and willing to publish and also fear committing errors and my head is already saturated with all the info... so I either:

    1) I keep doing as I have, and then selecting the depth level of my choice to show separately, as I have been doing ALSO.

    2) I get a slave to dye them by hand (just joking, of course), because I'm not going to do it.

    3) I get a script which does it even better and, in principle, with no chance of accidental error.

    This last is the real solution but I do not know (yet) how to do it. Ask me a year from now and maybe...

    ReplyDelete
  32. This comment has been removed by the author.

    ReplyDelete
  33. Algeria has important Berber groups who are relatively of pure Berber origin (Kabyles , Chawis , Chenouis , and other Zenatas of Western region) but none of them were studied or included in any studies

    ReplyDelete
  34. I'm working with what I have. Sorry that I can't improve or extend the samples. The data from Morocco in this case is also unspecific (north, south) and does not say if they are Berber, Arab or what.

    Instead the Tunisian sample (where we would expect less such a phenomenon maybe) is totally isolate. :(

    ReplyDelete
  35. Tunisia stands out as an isolate in this too

    http://dienekes.blogspot.com/2012/01/back-to-north-africa-henn-et-al-2012.html

    ReplyDelete
  36. "I'm working with what I have. Sorry that I can't improve or extend the samples. The data from Morocco in this case is also unspecific (north, south) and does not say if they are Berber, Arab or what. "
    nice job so far Maju , and I agree with your interpretations..

    the Northern Moroccan sample is most likely from a locality to the south of the city of Ksar-el-Kebir in the Gharb valley rather than the foothills of the Rif mountain range , it is an arabized region ....
    the "Northern" Algerian sample is not northern at all , it is from the Saharan Atlas region of Djelfa..not very representative of the majority coastal population..

    finally,I do think that the "Aterian" component in southern Moroccans is not exclusively Southern ,but, rather has a pan-Atlasian distribution from the Eastern Atlas (middle Atlas ) to the southwestern Anti-Atlas (Souss region )...

    ReplyDelete
  37. @Grey: It is strikingly the same exercise (only some details vary). However they produce some extra data worth commenting, notably they estimate times for West Asian and European genetic penetration that can well be coincident with Dabban "Aurignacoid" and Oranian "Solutroid" respectively. I'll address it soon (no time for everything some days).


    @Dalouh:

    "the Northern Moroccan sample is most likely from a locality to the south of the city of Ksar-el-Kebir in the Gharb valley rather than the foothills of the Rif mountain range , it is an arabized region ... the "Northern" Algerian sample is not northern at all , it is from the Saharan Atlas region of Djelfa..not very representative of the majority coastal population".

    How do you know?

    "finally,I do think that the "Aterian" component in southern Moroccans is not exclusively Southern ,but, rather has a pan-Atlasian distribution from the Eastern Atlas (middle Atlas ) to the southwestern Anti-Atlas (Souss region)"...

    Again how can you know? What's the foundation for this claim?

    If I'm going to accept your opinion (which may well be correct) the least I know is some objective support - or at the very least a well-sounding explanation.

    ReplyDelete
  38. Typo: "the least I know"... should read "the least I need"... or "I require"...

    ReplyDelete
  39. just by looking at a map like this one..

    http://img818.imageshack.us/img818/6426/henn2.jpg
    I have seen a clearer one elsewhere where the N-Moroccan sample points to Ksar-El-Kebir / Souk Larbaa Gharb area ...anyway, it doesn't make much difference if they speak Arabic or Berber in the Northern coastal region..
    the "Northern Algerian" sample, however, clearly points to the Saharan Atlas range,home of the Ouled Nail confederation , a semi-arid southern region..

    http://upload.wikimedia.org/wikipedia/commons/thumb/a/ab/Atlas-Mountains-Labeled-2.jpg/1024px-Atlas-Mountains-Labeled-2.jpg

    as for the "Aterian component", I am confident it is a pan-Atlasian component,I believe that the Moroccan highlanders (excluding the coastal Riffians ) were the last pre-Dabban Aterian populations to be absorbed by the Eurasians..
    from the Middle Atlas range :
    http://www.flickr.com/photos/jean-vigo/5922402570/

    http://www.flickr.com/photos/jamesgrassick/5053707989/sizes/z/in/photostream/

    http://www.flickr.com/photos/phil_marion/5533893102/

    ReplyDelete
  40. Alright, thanks. I must agree with the North Algerian evaluation as mountain Algeria. It is also interesting that the Tunisians are from inland sites, towards the desert, and not your typical coastal Tunisian at all. The North Moroccan however looks Northern enough to me, possibly from Tetouan (if the cross mark is accurate).

    "as for the "Aterian component", I am confident it is a pan-Atlasian component"

    It has really not shown up in the data but as minimal. Only South Moroccans showed it and that was in accordance with my expectations (from previous discussions on the "pseudo-Mogoloid" or "pseudo-Khoisanid" phenotype).

    I truth the phenotypes you posted are like confusing and at least two of them are very common in NW Africa. But how "Aterian" or how something else ("Ethiopid" for example: there are clear "East African" minority affinities in NW Africa) they are they is something I can't discern only by phenotype.

    ReplyDelete
  41. Dalouh fingered a good point. Anyway North Africa is diverse.
    In the North , you have the presence of types clearly related to Mediterranean and Middle-Eastern populations that represent the out-of-Africa migration pattern like in this galleries of photos :

    http://www.forumbiodiversity.com/showthread.php?t=13528&page=26
    http://www.forumbiodiversity.com/showthread.php?t=13528&page=28

    While you go much farther south , clear indigenous African features start to show up stronger like among Tuaregs : http://www.forumbiodiversity.com/showthread.php?t=15093

    Beautiful isn't it ^^.
    Genetics may would help us to understand the complicated various migration patterns.

    ReplyDelete
  42. The photos are very interesting, Ashtaga, even if a bit overwhelming for my computer's capacity. I'd say that there is a main North African type that here would be quickly identified as "Moorish". Then there is also a European type, often Iberian looking (but also "Russian looking" or "British looking" now and then), and all kind of intermediate types. In most cases the whiter (skin-tone) individuals look European (at least to my eye, but of course shade affects overall perception).

    The Moorish type is IMO Caucasoid (of mainly West Eurasian stock) but distinct because of the antiquity of divergence and maybe African admixture (incl. North African Aterian aboriginals).

    A particular clue I got from this woman, who is almost identical to famous Indian journalist Arundhati Roy. They are not the same person (look at tiny details for the differences) but you may be deceived easily. While Roy is obviously in the whitest (by skin tone) segment of Indian society, she's a pure-blooded Indian anyhow and it's almost impossible that the two women have any recent relation at all (and by 'recent' I mean post-Neolithic), what indicates how incredibly well can phenotypes be preserved, even if in randomized and fragmented form, across the millennia.

    ReplyDelete
  43. You welcome :) What you call as the 'Moorish" type I would say is the Berberid type.

    Good examples of relatively pure Berberid men and women
    http://www.forumbiodiversity.com/showpost.php?p=636689&postcount=249

    You are right the lighter variants are Iberian-like , likely remnants of the Mountainous Cro-Magnon Mechtoid :
    http://www.forumbiodiversity.com/showthread.php?t=13528&page=27
    http://www.forumbiodiversity.com/showthread.php?t=13528&page=24

    The woman you posted who ressemble the Indian Arudhati Roy is also a Berber type (looks like an older version of the French singer Naâdiya). There's definitely a continuum within human populations :)

    The French actor Tahar Rahim who's also Berber have got the small eyes related to that of those in the High Atlas mount

    ReplyDelete
  44. "What you call as the 'Moorish" type I would say is the Berberid type".

    Fair enough. Your term is surely more correct. I said 'Moorish' because that's what people in the street here would say: "parece moro": "he/she looks Moorish".

    What is interesting is that even if most of those who "look Moorish" are effectively North African by origin, not all North Africans "look Moorish" and many surely walk the streets of, say, Madrid, totally unnoticed (visually).

    "You are right the lighter variants are Iberian-like"

    Absolutely, your pictures are very clear for the trained eye (even if there are individuals who rather look like Russians or something else, they are rare).

    Something I'm chewing on is that Sahrawis are also distinctive on their own right. They certainly harbor no types who look European (apparent in the 'Admixture' analysis: 1.2% Iberian affinity only and) but specially, while most Berberids have brownish/olive skin tones, these have rather golden brown shades very dominantly. Have you noticed that?

    ReplyDelete
  45. Because they live in the Sahara I guess. Far from the Mediterranean to have an olive Mediterranean skintone , however they lack of European affinity is strange from what I have seen on NW Africa genetic. In fact on a a West Eurasian global plot , NW Africans cluster in between Egyptian/Arab and Sardinian populations. Their lack of European affinity is perhaps due to a higher SSA contribution to them?

    These people are from Medea (in central costal Algeria), funny that if you put dark hair and 1 shade darker to the reconstruction of the Mechta-Afalou reconstruction by Elizabeth , it will perfectly fit there :)
    http://anthrocivitas.net/forum/showthread.php?p=196571#post196571

    It would be really great if there was a genetic study on the Libyan Berbers as well , would they come out identical? (Libyan Arabs are quite divergent from the mass of NW Africans)
    http://www.forumbiodiversity.com/showpost.php?p=765479&postcount=1
    http://www.forumbiodiversity.com/showpost.php?p=766348&postcount=2
    http://www.forumbiodiversity.com/showpost.php?p=795511&postcount=11


    The must would be also that of the Tuareg people , they have very particular phenotypes , however I bet most are mixed , in fact even the Fulanis show significant Mediterranean/West Eurasian genes

    http://www.forumbiodiversity.com/showthread.php?t=15093&page=3

    http://www.forumbiodiversity.com/showthread.php?t=15093&page=3

    ReplyDelete
  46. Also the Chaoui people ( Berber ethnic group) , their data is in need :/ although we already got the schema general!
    http://www.forumbiodiversity.com/showpost.php?p=754714&postcount=305

    Btw I got something here!
    Chromosomes of a Rifian : http://i1086.photobucket.com/albums/j442/mimimerabet/M030051_8E4F6B.png?t=1332852831
    Of a Kabyle:http://img804.imageshack.us/img804/8829/biloomotherk12b.jpg

    ReplyDelete
  47. "Their lack of European affinity is perhaps due to a higher SSA contribution to them?"

    No. This is relative and you can see, with great accuracy, the West Eurasian (light blue) and Tropical African components in the K=2.

    There's always some blurriness and uncertainty because all that the algorithm does is to split the sample in K groups (two for our discussion right now) and say these belong here by X% and ther by Y%. In our example K=2 purest populations are made to be Spaniards and Mandenka and all the rest have a bit of each but mostly of the blue component (except Fulani and to some extent Ethiopians and South Moroccans).

    Sahrawis are rather high in the blue component, just like (Central) Moroccans.

    The case of South Moroccans is probably misleading because they have (I found in this very exercise) a minor component that is very distant from both the West Eurasian and the Tropical African populations. This surely gets divided 50-50 at the K=2 and only shows up at K=11.

    As for phenotypes... right now I find them rather mind-boggling. So I'll better don't say anything else.

    ReplyDelete

Please, be reasonably respectful when making comments. I do not tolerate in particular sexism, racism nor homophobia. Personal attacks, manipulation and trolling are also very much unwelcome here.The author reserves the right to delete any abusive comment.

Preliminary comment moderation is... ON (sorry, too many trolls).