tag:blogger.com,1999:blog-7785493.post8888652388251653386..comments2024-01-04T04:11:55.717+02:00Comments on Dienekes’ Anthropology Blog: A surprising link between Africans and DenisovansDienekeshttp://www.blogger.com/profile/02082684850093948970noreply@blogger.comBlogger29125tag:blogger.com,1999:blog-7785493.post-23847016834625996872012-10-01T04:25:28.812+03:002012-10-01T04:25:28.812+03:00Something I've just realised is rather interes...Something I've just realised is rather interesting. The Denisova element in Africa tends to be found amoung Mbuti, Yoruba and Mandenka. These groups are found very near the region where A0 and A1 are supposed to have first split. Perhaps the Denisova element lies at the beginning of the modern Y-DNA line. Perhaps the Denisova element was then diluted by admixture with whatever population supplied the modern mt-DNA line. If that is so then hybridism lies right at the base of the origin of 'modern' humans in Africa. Certainly the mtDNA L0/L1''6 split looks to lie somewhat east of the Y-DNA A0/A1 split. terrythttps://www.blogger.com/profile/17327062321100035888noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-54452130320342997892012-09-30T14:00:35.501+03:002012-09-30T14:00:35.501+03:00Yes, I just read the 3-page Affymetrix .pdf - trul...Yes, I just read the 3-page Affymetrix .pdf - truly a good way to derive SNPs. The only possible bias I can think of is from panel 13. For example, Denisovans may have reversed to Chimp from all other human lineages, there, and a deviation from most humans may indicate Denisovan admixture rather than a San mutation.eurologisthttps://www.blogger.com/profile/03440019181278830033noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-10651940191919362152012-09-30T13:22:06.807+03:002012-09-30T13:22:06.807+03:00Tobus,
I understand now - I think I simply misrea...Tobus,<br /><br />I understand now - I think I simply misread what you stated before. eurologisthttps://www.blogger.com/profile/03440019181278830033noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-23823297928531671242012-09-30T09:18:17.381+03:002012-09-30T09:18:17.381+03:00Also, outgroups are not needed to define SNPs, the...Also, outgroups are not needed to define SNPs, they are only useful to infer their polarity, e.g., whether there was a G->T or T->G mutation.Dienekeshttps://www.blogger.com/profile/02082684850093948970noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-31806290598872356652012-09-30T09:15:12.739+03:002012-09-30T09:15:12.739+03:00Human SNPs are defined by using an outgroup, like ...<i>Human SNPs are defined by using an outgroup, like chimps or San. Pygmy's and non-Africans could perfectly be monomorph for a certain SNP, in that case they group together.</i><br /><br />The SNPs in the Affymetrix Human Origins Array were discovered in two chromosomes of single human individuals. Even if there were such sites included, they would make no difference for D-statistics, because D-statistics only count BABA and ABBA sites, i.e., sites where modern human populations differ from each other.<br /><br />Also, positions where modern humans have fixed differences from outgroups are not SNPs. SNPs are by definition polymorphisms, and fixed sites within a species are not polymorphisms. Such sites are useless for purposes of studying human population differences, because, obviously, populations don't differ in them.Dienekeshttps://www.blogger.com/profile/02082684850093948970noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-89873812817068788882012-09-29T23:47:33.091+03:002012-09-29T23:47:33.091+03:00"If they're monomorphic in both Africans ..."If they're monomorphic in both Africans and Eurasians, then they're not even SNPs"<br /><br />Human SNPs are defined by using an outgroup, like chimps or San. Pygmy's and non-Africans could perfectly be monomorph for a certain SNP, in that case they group together.Rokushttps://www.blogger.com/profile/13883125231922541439noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-14748650891845605582012-09-29T16:23:33.219+03:002012-09-29T16:23:33.219+03:00but since they are markedly different to the ALL p...<i>but since they are markedly different to the ALL panel results (4% Denisovan for Papuans, nothing significant for Mbuti) I suspect the GLOBAL panel has some inherent bias as well </i><br /><br />The difference between the ALL panel and the others, is that the ALL panel includes SNPs that are polymorhpic in Africans and monomorphic in Eurasians. So, the signal disappears because of the archaic admixture in Africans.<br /><br />The signal is also present in SNPs ascertained in a single Papuan individual, as shown in the newer post.Dienekeshttps://www.blogger.com/profile/02082684850093948970noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-30027656593397159472012-09-29T15:53:05.988+03:002012-09-29T15:53:05.988+03:00Well, I guess I still don't get it. If a site ...<i>Well, I guess I still don't get it. If a site is monomorphic in one population, but polymorphic in another, I have no idea how the D-statistic knows that the monomorphic one is non-Chimp (and not vice-versa). Naively, I would think the opposite.</i><br /><br />Perhaps if you tell me why you think the monomorphic allele would be Chimp I can explain how I see it differently.... so why do you think the monomorphic one is more likely to be Chimp?Tobushttps://www.blogger.com/profile/05529220083970625733noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-71024208826494045412012-09-29T15:47:43.807+03:002012-09-29T15:47:43.807+03:00Look at the Z-GLOBAL column as well. This involves...<i>Look at the Z-GLOBAL column as well. This involves SNPs that are polymorphic in ALL 5 major human groups. MbutiPygmy are significantly more "Denisovan" in that SNP set as well.</i><br /><br />You are correct - the Mbuti are showing similar results to the Papuans for the GLOBAL panel (1.5-2% more Denisovan). There may be something significant about these results, but since they are markedly different to the ALL panel results (4% Denisovan for Papuans, nothing significant for Mbuti) I suspect the GLOBAL panel has some inherent bias as well - impossible to tell why without more information of how it's made up, but since "polymorphic" can mean both "90% A, 10 %B" and "10% A, 90% B", I would guess that the allele frequencies are not consistent across the populations.Tobushttps://www.blogger.com/profile/05529220083970625733noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-365652555339821362012-09-29T15:21:46.082+03:002012-09-29T15:21:46.082+03:00The majority of monomorphic alleles will be non-Ch...<i>The majority of monomorphic alleles will be non-Chimp because the D-statistic only selects those alleles that have mutated at least once since the Chimp/Homo divergence</i><br /><br />Well, I guess I still don't get it. If a site is monomorphic in one population, but polymorphic in another, I have no idea how the D-statistic knows that the monomorphic one is non-Chimp (and not vice-versa). Naively, I would think the opposite. eurologisthttps://www.blogger.com/profile/03440019181278830033noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-26625965601548641672012-09-29T14:53:07.911+03:002012-09-29T14:53:07.911+03:00Pardon my ignorance - am I missing something that ...<i>Pardon my ignorance - am I missing something that makes the monomorphic places automatically non-Chimp?</i><br /><br />The majority of monomorphic alleles will be non-Chimp because the D-statistic only selects those alleles that have mutated at least once since the Chimp/Homo divergence - it only counts ABBA and BABA combinations so the alleles must be different between Denisovan and Chimps for the D-statistic to count them. Using 6Mya as the Chimp/Homo divergence and 1Mya for Denisovan/Sapiens, assuming a constant mutation rate we would expect 5My worth or 83% of the non-Chimp mutations occuring in the mutual Denisovan/Sapiens ancestor. So the vast majority of SNPs being counted will have the Denisovan allele as the "normal" one for Pop1 and Pop2, with the Chimp allele only occuring due to mutation or admixture. Since polymorphic alleles are more likely to be derived (at least twice as likely given the very conservative 33% chance each of drift, mutation or admixture causing the polymorphism), then the majority of the monomorphic alleles will be Denisovan.<br /><br /><br /><i>Moreover, the polymorphic ones are much more likely to be a single than a double mutation, in the first place.</i><br /><br />I'm not sure exactly what you mean by "single" and "double" mutation, but if you mean the number of times it's mutated from the Chimp allele, then all the SNPs will be at least "single" (Chimp to Denisovan) and the polymorphic ones more likely to be "double" (Chimp to Denisovan/Sapiens ancestor to Eurasian).<br />Tobushttps://www.blogger.com/profile/05529220083970625733noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-42675435224886741652012-09-29T09:10:39.991+03:002012-09-29T09:10:39.991+03:00@Tobus,
Look at the Z-GLOBAL column as well. This...@Tobus,<br /><br />Look at the Z-GLOBAL column as well. This involves SNPs that are polymorphic in ALL 5 major human groups. MbutiPygmy are significantly more "Denisovan" in that SNP set as well.Dienekeshttps://www.blogger.com/profile/02082684850093948970noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-17524197404550192792012-09-29T08:09:36.477+03:002012-09-29T08:09:36.477+03:00Tobus,
Wouldn't the polymorphic European SNPs...<i>Tobus,</i><br /><br />Wouldn't the polymorphic European SNPs have an even <i>higher</i> chance of being <i>away</i> from Chimp (any mutation if the monomorphic African is Chimp-like) than <i>towards</i> chimp (only 1/3, in case the African is non-Chimp-like)? Moreover, the polymorphic ones are much more likely to be a single than a double mutation, in the first place.<br /><br />Pardon my ignorance - am I missing something that makes the monomorphic places automatically non-Chimp?eurologisthttps://www.blogger.com/profile/03440019181278830033noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-10963744394199101272012-09-29T05:11:32.710+03:002012-09-29T05:11:32.710+03:00"Put another way, archaic admixture at say 20..."Put another way, archaic admixture at say 200 kya is effectively parts of the definition of the AMH species, and hence, really isn't admixture at all". <br /><br />I agree that admixture has been a factor in our evolution right from before the appearance of Australopithecus, but we're not dealing here with that kind of ancient admixture. <br /><br />"So, something different is going on with respect to the relationship of Africans to Neandertals/Denisovans". <br /><br />Easily solved if we accept the possibility that at least the modern mt-DNA owes its deeper origins to somewhere outside Africa. Surely there is absolutely no reason to blindly accept that 'modern' humans evolved from an African population whose ancestors had remained behind in Africa, and they had always been confined to that continent. <br /><br />"There are 3 ways this polymorphic/monomorphic split can come about: an allele was lost due to selection/drift in the monomorphic population, a new allele emerged due to mutation in the polymorphic population, or a new allele was introgressed from a 3rd population". <br /><br />Good point. terrythttps://www.blogger.com/profile/17327062321100035888noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-42314449260670434482012-09-29T03:10:32.210+03:002012-09-29T03:10:32.210+03:00Look at the data again.
In both the AFRICA and EUR...<i>Look at the data again.<br />In both the AFRICA and EURASIA panel, Africans are less Neandertal than Eurasians.<br />In the AFRICA panel they are less Denisovan, and in the EURASIA panel they are more Denisovan</i><br /><br />I have looked at the data again and I can see what you are saying... Eurasians are scoring -10% or so for Denisovans when compared to Africans using the polymorphic Eurasian SNPs. We can see for Neandertal admixture Eurasians are scoring 33% and 12-15% for the AFRICA and EURASIA panels, while for Denisovan they are scoring 25% and -8-10% respectively. This -10% for Denisovan stands out remarkably and indicates hitherto unknown Denisovan admixture in African populations.<br /><br />However, I have shown that there is a minimum 10% expected bias against a population when testing against a set of SNPs that only includes polymorphisms for that population. This bias is because 1 out of every 3 times that the SNP is caused by a genetic mutation in the population, it will revert to the ancestral allele and thus look like an affinity when in fact it isn't. If we allow for a 10% inflation of the scores then we get Neandertal admixture in Eurasians scoring 23% and 22-25% for the AFRICA and EURASIA panels and Denisovan admixture gets 12% and 0-2% respectively. Pretty much in line with expectations.<br /><br />The -ve Eurasian/Denisovan value for the EURASIA panel stands out, but it can be accounted for as the result of inherent bias in the polymorphic/monomorphic SNP selection. The "something different ... going on with respect to the relationship of Africans to Neandertals/Denisovans" is probably just a side effect of the method you are using.Tobushttps://www.blogger.com/profile/05529220083970625733noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-35110720213865541002012-09-29T02:54:35.609+03:002012-09-29T02:54:35.609+03:00Dienekes, do you know the approximate date when th...Dienekes, do you know the approximate date when this admixture could've occurred?<br /><br />I think the second explanation, an into-Africa migration, is more plausible.<br /><br />Or, the archaic human group found by Michael Hammer and Joseph Lachance was already akin to the Denisovans.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-7785493.post-46483839965715486332012-09-28T22:12:54.036+03:002012-09-28T22:12:54.036+03:00Put another way, archaic admixture at say 200 kya ...<i>Put another way, archaic admixture at say 200 kya is effectively parts of the definition of the AMH species, and hence, really isn't admixture at all.</i><br /><br />You are confusing the time when admixture happened, with the time when the archaic population split. For example, the archaic admixture that was detected by Hammer et al. and Lachance et al. happened a few tens of kya, but involved hominins that split off from modern humans hundreds of kya.<br /><br />Dienekeshttps://www.blogger.com/profile/02082684850093948970noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-30446869063906809162012-09-28T21:55:42.212+03:002012-09-28T21:55:42.212+03:00"At longer time depths (say before 100ka) arc..."At longer time depths (say before 100ka) archaic segments are so small that it becomes probabilistically difficult to show that they are archaic."<br /><br />At time depth >> 100kya, you are very close to the root of the modern human tree and predate any divergence of populations ancestral to modern ones into separate population genetic populations.<br /><br />Put another way, archaic admixture at say 200 kya is effectively parts of the definition of the AMH species, and hence, really isn't admixture at all.andrewhttps://www.blogger.com/profile/08172964121659914379noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-80298738800796753362012-09-28T17:19:11.712+03:002012-09-28T17:19:11.712+03:00But it does occur for Neandertals
Look at the dat...<i>But it does occur for Neandertals</i><br /><br />Look at the data again.<br />In both the AFRICA and EURASIA panel, Africans are less Neandertal than Eurasians.<br />In the AFRICA panel they are less Denisovan, and in the EURASIA panel they are more Denisovan (except when compared to Papuans who beat everyone else in the Denisovan department).<br /><br />So, something different is going on with respect to the relationship of Africans to Neandertals/Denisovans.<br />Dienekeshttps://www.blogger.com/profile/02082684850093948970noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-40577757622000196132012-09-28T16:32:41.528+03:002012-09-28T16:32:41.528+03:00>I'm not sure this isn't a side-effect ...<i>>I'm not sure this isn't a side-effect of using only polymorphic Eurasian allelles that are monomorphic in the Mbuti.<br /><br />It does not occur for the equivalent comparisons for Neandertals, so something specific is going on for Denisovans.</i><br /><br />But it does occur for Neandertals, as you noted: "Eurasians appear substantially Neandertal/Denisovan-admixed when SNPs polymorphic in Africans and monomorphic in Eurasians are used. I can think of no other explanation than archaic African admixture for this finding." While archaic admixture in Africa is a definite possibility, this finding is exactly what we'd expect to see if using a set of SNP's that is weighted towards African mutations inflates the apparent Chimp affinity for African populations. It's worth noting that the D-statistics for these SNP's with the Mbuti are much higher than for any other results in the table - 33% (Neandertal) and 23% (Denisovan), which would support the idea an inflated result. <br /><br />Following this logic, the 12-14% result for Eurasian vs Mbuti Neandertal using the Eurasian SNP's would be underestimated.Tobushttps://www.blogger.com/profile/05529220083970625733noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-25922215855873854572012-09-28T13:33:46.641+03:002012-09-28T13:33:46.641+03:00Great, but by excluding non-African monomorphic SN...<i>Great, but by excluding non-African monomorphic SNPs you also exclude the possibility that most of these SNPs could be ancestral.</i><br /><br />If they're monomorphic in both Africans and Eurasians, then they're not even SNPs.Dienekeshttps://www.blogger.com/profile/02082684850093948970noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-51796070818645942072012-09-28T13:10:04.456+03:002012-09-28T13:10:04.456+03:00"EURASIA: 94858 SNPs that were polymorphic in..."EURASIA: 94858 SNPs that were polymorphic in at least one non-African population and monomorphic in MbutiPygmy"<br /><br />Great, but by excluding non-African monomorphic SNPs you also exclude the possibility that most of these SNPs could be ancestral. In that case, you make the ancestral part of Eurasian SNPs disappear and create a marvellous homozygotous correlation with Africans instead.Rokushttps://www.blogger.com/profile/13883125231922541439noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-71654365775707438242012-09-28T12:52:09.394+03:002012-09-28T12:52:09.394+03:00There have been a couple of studies that have used...<i>There have been a couple of studies that have used statistical methods to show some archaic admixture in Paleoafricans, although not a whole lot (maybe 1%). The same statistical methods predicted Neanderthal admixture and Denisovian admixture as well (although didn't know what to make of it at the time).</i><br /><br />Because of the absence of archaic African genomes, inferrence of archaic African admixture is limited to a recent time depth. At longer time depths (say before 100ka) archaic segments are so small that it becomes probabilistically difficult to show that they are archaic.<br /><br />It's like throwing a bad coin 5 times vs. 100. At 100 times, it's likely that you'll figure out a deviation from 50-50 chance of heads/tails. At 5 times, it's less likely.<br /><br />The occurrence of substantial African-specific polymorphism can be explained either in terms of a massive bottleneck during OoA or due to admixture with divergent African populations which contributed new polymorphism. A bottleneck does not explain the pattern of the D-statistics, because it would imply that in the random game of chance, Out-of-Africans ended up with the same alleles as Neandertals, which does not seem plausible.Dienekeshttps://www.blogger.com/profile/02082684850093948970noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-74485569035954574052012-09-28T12:44:59.633+03:002012-09-28T12:44:59.633+03:00I'm not sure this isn't a side-effect of u...<i>I'm not sure this isn't a side-effect of using only polymorphic Eurasian allelles that are monomorphic in the Mbuti.</i><br /><br />It does not occur for the equivalent comparisons for Neandertals, so something specific is going on for Denisovans.Dienekeshttps://www.blogger.com/profile/02082684850093948970noreply@blogger.comtag:blogger.com,1999:blog-7785493.post-10492844887148639072012-09-28T11:36:56.017+03:002012-09-28T11:36:56.017+03:00There have been a couple of studies that have used...<i>There have been a couple of studies that have used statistical methods to show some archaic admixture in Paleoafricans, although not a whole lot (maybe 1%).</i><br /><br />Such a low archaic admixture in Paleoafricans is increasingly looking less and less likely with new studies. According to the most recent studies, Paleoafricans probably have the highest amounts of archaic admixture among all existing humans.Onur Dincerhttps://www.blogger.com/profile/05041378853428912894noreply@blogger.com