The evolution of skin pigmentation-associated variation in West Eurasia.

I have run La Brana and several other higher coverage WHG through on the calculator, they all get lighter skin predictions when you swap the blue eye associated variants for the ancestral brown eye alleles. I cannot post screenshots but if you want to put the data through yourself here is the number of alleles to input for each snp for La Brana for the darker prediction range suggested in the original study:

12: 2
13: 2
16: 1
19: 1
27: 2
29: 1
35: 2
39: 1

Then see how the skin prediction changes when you plug in 2 alleles for snps:

20, 31 & 32

You will notice that whilst the blue eyes are totally gone, the skin prediction changes to dark from dark to black.

So, academics putting out a major paper just decided not to input the snps for eye color???
 
I have run La Brana and several other higher coverage WHG through on the calculator, they all get lighter skin predictions when you swap the blue eye associated variants for the ancestral brown eye alleles. I cannot post screenshots but if you want to put the data through yourself here is the number of alleles to input for each snp for La Brana for the darker prediction range suggested in the original study:

12: 2
13: 2
16: 1
19: 1
27: 2
29: 1
35: 2
39: 1

Then see how the skin prediction changes when you plug in 2 alleles for snps:

20, 31 & 32

You will notice that whilst the blue eyes are totally gone, the skin prediction changes to dark from dark to black.

did you enter NA for all the others?
 
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,
 
Last edited:
All of the other snps you leave as '0' with no inputs as 'NA'

is that really correct? we don't know the data for those snp's right? i assume you are using this right? https://hirisplex.erasmusmc.nl/

i already wrote this somewhere else but i don't get that site.

for example for rs1129038(31) it says "G" but when i google that snp it says that there is a C and a T allele not G
 
Have a look on snpedia and search for 'orientation' - in this instance the C/G allele represents the ancestral brown eye associated variant, and the T/A allele represents the derived associated blue eye variant at highest frequency to North Europe

thanks. i thought that it might be because of the reversed/non-reversed strand. but didn't think people would actually label SNP's based on the non-coding strand or mix everything up.

tried the different snp's and it seems for the different HERC2 variants nr.20, 28, 30 leads to darker skin prediction while 31 and 32 lead to a lighter one. it's a bit strange but i'm not sure how accurate this is if you just put in 0 in those where we don't know the value.

something that might be a sign of this problem that you really shouldn't just add 0 everywhere: if you leave everything at 0 and just change 31 and 32 then they actually lead to somewhat darker skin prediction while before with the values of la brana they lead to lighter skin prediction.
 
........................
 
Last edited:
Yes 20 and 28 are giving what I would consider a consistent effect as with modern populations, 31 and 32 are inconsistent with modern populations - especially 31 - and the derived allele is almost unique to North Europe and does not appear anywhere else at any great frequencies bar the Americas of course.

So what you have here is alleles most unique to North Europe contributing directly to heightened likelihood of black skin...

Now see what happens when applying the same conditions to this modern day Finnish prediction; inputs are:

12: 1
13: 1
22: 1
24: 1
25: 1

All other snps are left as '0'



And also Loschbour:

12: 1
13: 2
15: 2
16: 2
27: 2
29:2
35: 2


for those predictions ancestral version of 31 and 32 lead to higher likelyhood of dark skin. isn't that what you would expect?

imo it shows that it is highly dependant on the context and that it's probably not accurate if we put in 0 everywhere instead of NA.
 
Yes that is my point. The Loschbour WHG prediction of 0.90 for intermediate is consistent - so 31 and 32 ancestral give the darkening effect to the prediction just like the modern day Finnish.

However with La Brana the identical ancestral alleles at highest frequency to Africa/ East Asia contribute significantly to a lighter prediction. Remember these HGs were relatively homogenous, so any factors such as epistasis or variants not included in the model that could be influencing skin pigmentation on the hirisplex-s database subjects is not likely to be relevant to such similar individuals.

We have to also remember that there is not a single actual WHG on the hirisplex database - but, of all potential comparative subjects, would it be more relevant to compare a WHG to say a Finnish person, or a heavily mixed African American - of which there will be numerous on the hirisplex database? I should also mention that on the snipper skin prediction model, all WHG score intermediate - Loschbour and La Brana actually score identical on the snipper.

can you link me to the allele table for the WHG's? are those values you put in 0 really not there or are they NA?
 
Yes that is my point. The Loschbour WHG prediction of 0.90 for intermediate is consistent - so 31 and 32 ancestral give the darkening effect to the prediction just like the modern day Finnish.

However with La Brana the identical ancestral alleles at highest frequency to Africa/ East Asia contribute significantly to a lighter prediction. Remember these HGs were relatively homogenous, so any factors such as epistasis or variants not included in the model that could be influencing skin pigmentation on the hirisplex-s database subjects is not likely to be relevant to such similar individuals.

We have to also remember that there is not a single actual WHG on the hirisplex database - but, of all potential comparative subjects, would it be more relevant to compare a WHG to say a Finnish person, or a heavily mixed African American - of which there will be numerous on the hirisplex database? I should also mention that on the snipper skin prediction model, all WHG score intermediate - Loschbour and La Brana actually score identical on the snipper.


i tried Snipper now too and i have one question. in Snipper rs16891982 is C among 84% of "whites" so i assume if you enter C in Snipper for that snp it will give you a lighter skin prediction. however i believe the C from Hirisplex is actually not the C from Snipper because if you google that snp the ancestral allele is C and there are 2 possible mutations either A or G which means depending on which strand people look at, in case of the G muation couldn't it be that this mutation and the ancestral allele potentially get confused for each other?
in that case i would enter G instead of C in snipper for that SNP and then you will get intermediate and black prediction.
 
i tried Snipper now too and i have one question. in Snipper rs16891982 is C among 84% of "whites" so i assume if you enter C in Snipper for that snp it will give you a lighter skin prediction. however i believe the C from Hirisplex is actually not the C from Snipper because if you google that snp the ancestral allele is C and there are 2 possible mutations either A or G which means depending on which strand people look at, in case of the G muation couldn't it be that this mutation and the ancestral allele potentially get confused for each other?
in that case i would enter G instead of C in snipper for that SNP and then you will get intermediate and black prediction.


i checked again, the variant most common in europe is derived not ancestral, which means you probably really have to enter G instead of C in Snipper which will yield an intermediate to black skin prediction.
 
.........................
 
Last edited:
Yes C would represent the derived allele on the snipper, also rs13289 the derived allele you input C rather than G

What I found more puzzling than the prediction of the Cheddar man being "dark dot black" was the phenotype analysis for the BA Aegeans. HIrisPLex predicted 1 BA Aegean individual to be in the "dark" category while the other two were predicted to be most likely "dark to black". These 3 individuals are AA on rs1426654 (SLC24A5) but CC and CG (overwhelmingly GG in contemporary Europeans) on rsrs16891982 (SLC45A2). With that being said, the two Aegeans that scored "dark to black" carried OCA2 which also contributes to pigmentation. How realistic do you find the phenotype analysis in terms of BA Greeks having dark to black skin tone?
 
What I found more puzzling than the prediction of the Cheddar man being "dark dot black" was the phenotype analysis for the BA Aegeans. HIrisPLex predicted 1 BA Aegean individual to be in the "dark" category while the other two were predicted to be most likely "dark to black". These 3 individuals are AA on rs1426654 (SLC24A5) but CC and CG (overwhelmingly GG in contemporary Europeans) on rsrs16891982 (SLC45A2). With that being said, the two Aegeans that scored "dark to black" carried OCA2 which also contributes to pigmentation. How realistic do you find the phenotype analysis in terms of BA Greeks having dark to black skin tone?

Wow that is some interesting information. The AA alleles SNP rs1426654 has been shown to explain 25% to 38% of the variation in skin tone between Europeans and West Africans, etc. (I have the paper in my files somewhere). The rs16891982 SNP on SLC45A2 does impact skin tone, but I don't think as much as SLC24A5.

The US Department of Justice (USDOJ) published a study on genetics and skin tone which showed that 3 single polymorphisms on SLC24A5, SLC45A2 and ASIP in a Multiple Regression analysis explain 45.6% of skin tone variation across populations. When they added an Interaction term (ASIP*SLC45A2) to the model, they get 49.6% of the skin tone variation.

So it would be something like Y (Skin Tone) = Constant + B1 (SLC24A5) + B2 (SLC45A2) + B3 (ASIP) + B4 (SLC45A2*ASIP).....and other variables + error term.

USDOJ Final Copy of Research Paper conducted by Academicians at the University of Arizona (USA).

https://www.ojp.gov/pdffiles1/nij/grants/223980.pdf

The 2 SNP's in ASIP that impact Skin tone are rs6058017 and rs2424984. Both were significant in an ANOVA approach but the rs2424984 gave a slightly better fit in the Multiple Linear Regression analysis

From the USDOJ study (page 10):

"Using MLR, we found that both SNPs rs1426654 (SLC24A5) and rs16891982 (SLC45A2) described much of the variation in skin pigmentation across populations. The third most significant genetic contributor in a three-SNP MLR model was SNP rs2424984 (ASIP). ASIP has been shown to be associated with skin pigmentation, namely for rs6058017 (BONILLA et al. 2005; KANETSKY et al. 2002). Although we found rs6058017 to be significant by ANOVA, we did not find it to be a better predictor in skin reflectance than rs2424984. For both a single SNP analysis and a three-SNP model, rs2424984 was a better predictor for skin reflectance."
 
What I found more puzzling than the prediction of the Cheddar man being "dark dot black" was the phenotype analysis for the BA Aegeans. HIrisPLex predicted 1 BA Aegean individual to be in the "dark" category while the other two were predicted to be most likely "dark to black". These 3 individuals are AA on rs1426654 (SLC24A5) but CC and CG (overwhelmingly GG in contemporary Europeans) on rsrs16891982 (SLC45A2). With that being said, the two Aegeans that scored "dark to black" carried OCA2 which also contributes to pigmentation. How realistic do you find the phenotype analysis in terms of BA Greeks having dark to black skin tone?

looking at the frequencies of those genes in modern populations i would guess they were probably intermediate. but those are only very few genes. for example the derived allele of SLC24A5 is at more than 50% frequency in ethiopians and somalis but they aren't really light skinned. i imagine them to have had the skin complexion of modern day north african, levantine populations. those are also almost fixated for SLC24A5. Krause said somewhere that the farmers in europe had the complexion of modern populations from around the mediterranean so the same probably applies to bronze age aegeans.
 

This thread has been viewed 34356 times.

Back
Top