Page 1 of 4 123 ... LastLast
Results 1 to 10 of 37

Thread: Denisovan Analysis

  1. #1
    Banned
    Posts
    5,171
    Sex

    Lightbulb Denisovan Analysis

    I had always wanted to do Denisovan comparisons with moderns, but ascertainment bias had always rendered the results of any such comparisons with moderns pretty much useless. As a result, I could never do any meaningful IBS comparisons, or put together a decent admixture calculator using Denisovans. Analysis using the the normal datasets in use were always giving weird results such as Europeans and Africans showing more similarity to Denisovans.

    I dexided to put some time aside to see if I could overcome the problem with ascertainment bias, which is basically a problem of the SNPs sampled not being representative of the alleles in an individual. For example, the widely used Eurasian ascertained arrays, which are based on polymorphic loci that distinguish Eurasians, do not permit accurate comparisons with others with widely divergent histories from Eurasians, such as Africans, and archaics, such as Denisovan. Therefore the Eurasian ascertained arrays don't properly represent derived alleles uniquely attributed to Denisovans.

    It appears that I now finally have a decent dataset that I can use as a backbone for future Denisovan comparisons. This may lead to the 1st decent GEDmatch Denisovan calculator. Here are some IBS results, sorted with highest similarity to Denisovans. The results seem quite reasonable, especially since I am not using whole genome data, as would be expected to get meaningful results for any such comparisons.

    Immediately apparent, is Papuans, and other Austronesians scoring much higher similarity to Denisovan. I was also pleased to see that E Asians were on top and not in the bottom, and Africans in the bottom. All in all good news.


    NO SAMPLE / POP DENISOVA1 IBS
    1 Denisova_light 100.00%
    2 Denisovan 98.74%
    3 Vindija_light 84.17%
    4 Australian 62.42%
    5 Papuan 62.40%
    6 Bougainville 62.22%
    7 MA1 61.84%
    8 Itelmen 61.55%
    9 Eskimo 61.54%
    10 Chukchi 61.54%
    11 Nganasan 61.50%
    12 Karitiana 61.49%
    13 Cambodian 61.46%
    14 Ami 61.44%
    15 Dai 61.44%
    16 GujaratiD 61.43%
    17 Kusunda 61.42%
    18 Japanese 61.42%
    19 Han 61.42%
    20 Mongola 61.41%
    21 GujaratiC 61.36%
    22 Uzbek 61.35%
    23 GujaratiB 61.33%
    24 Scottish 61.32%
    25 Kalash 61.31%
    26 GujaratiA 61.29%
    27 Bengali 61.29%
    28 English 61.29%
    29 Estonian 61.28%
    30 Norwegian 61.28%
    31 Balochi 61.27%
    32 Iranian 61.27%
    33 Greek 61.26%
    34 Loschbour 61.26%
    35 Pathan 61.26%
    36 Algerian 61.23%
    37 Sardinian 61.23%
    38 LBK 61.23%
    39 Spanish 61.22%
    40 Saudi 61.22%
    41 Egyptian 61.19%
    42 Armenian 61.17%
    43 Somali 61.16%
    44 Hadza 61.09%
    45 Khomani 58.72%

  2. The Following 19 Users Say Thank You to Kurd For This Useful Post:

     Dinaek (01-18-2016),  evon (01-18-2016),  Hanna (01-18-2016),  Kaido (01-18-2016),  Kale (01-18-2016),  khanabadoshi (01-18-2016),  Megalophias (01-18-2016),  MonkeyDLuffy (01-18-2016),  N21163 (01-30-2016),  NK19191 (01-18-2016),  parasar (01-18-2016),  Passa (01-18-2016),  paulgill (01-18-2016),  Piquerobi (01-18-2016),  psaglav (01-18-2016),  surbakhunWeesste (01-23-2016),  Táltos (01-18-2016),  Tomenable (01-21-2016),  Varun R (01-18-2016)

  3. #2
    Banned
    Posts
    5,171
    Sex

    I will be posting some Dstats using my new dataset later. They tend to corroborate the IBS results

  4. The Following 2 Users Say Thank You to Kurd For This Useful Post:

     khanabadoshi (01-18-2016),  paulgill (01-18-2016)

  5. #3
    Registered Users
    Posts
    1,742
    Sex
    Omitted

    Good job Kurd!

    Look at MA1 on that list, right after Oceanian populations! I don't know whether to think that might be local stuff (I mean MA1 is just a few hundred kilometers from Denisova cave)...or whether there is some actual interaction with proto-Oceanian populations that some people seemed to eek out of stats. Whatever it was it looks like there was some continuity because next after MA1 are all Siberian populations.

  6. The Following 3 Users Say Thank You to Kale For This Useful Post:

     khanabadoshi (01-18-2016),  Kurd (01-18-2016),  NK19191 (01-18-2016)

  7. #4
    Banned
    Posts
    5,171
    Sex

    Quote Originally Posted by Kale View Post
    Good job Kurd!

    Look at MA1 on that list, right after Oceanian populations! I don't know whether to think that might be local stuff (I mean MA1 is just a few hundred kilometers from Denisova cave)...or whether there is some actual interaction with proto-Oceanian populations that some people seemed to eek out of stats. Whatever it was it looks like there was some continuity because next after MA1 are all Siberian populations.
    FWIW, MA1 was the only sample with a genotype rate of about 80%, which may have negatively impacted his score. The rest of the samples were at about 95% genotype rate

  8. The Following 4 Users Say Thank You to Kurd For This Useful Post:

     khanabadoshi (01-18-2016),  NK19191 (01-18-2016),  paulgill (01-18-2016),  Varun R (01-18-2016)

  9. #5
    Banned
    Posts
    5,171
    Sex

    UPDATE

    I will be posting Dstats as soon as I free up from ADMIXTURE. I have done no less than 20 ADMIXTURE runs on the RSID converted, V4 pruned, Denisovan master dataset, including project members. Right now the only issue seems to be some overlapping allele frequencies between Africans and Archaics at certain loci. While not so much of an issue with Dstats and IBS, it is one with ADMIXTURE, which is more sensitive to allele frequency matches on portions of the genome.

    I don't believe that Archaics and Africans necessarily share ancestry at those sites to the exclusion of Eurasians. I believe that those are simply ancestral sites where Eurasians have a derived allele for whatever reason (drift, bottlenecks, etc). In any case I have finally identified those problematic loci, and removed them from the analysis, while at the same time preserved those other loci where Archaics have elevated allele frequencies to the exclusion of Africans.
    Last edited by Kurd; 01-18-2016 at 04:18 PM.

  10. The Following 5 Users Say Thank You to Kurd For This Useful Post:

     Hanna (01-18-2016),  khanabadoshi (01-19-2016),  kingjohn (01-18-2016),  paulgill (01-18-2016),  Varun R (01-18-2016)

  11. #6
    Gold Class Member
    Posts
    7,801

    Quote Originally Posted by Kurd View Post
    FWIW, MA1 was the only sample with a genotype rate of about 80%, which may have negatively impacted his score. The rest of the samples were at about 95% genotype rate
    Were Ust-Ishim and the Denisova Neanderthal in this comparison?
    Also the Brahui and the Makrani, where do they fall?
    Last edited by parasar; 01-18-2016 at 05:10 PM.

  12. The Following User Says Thank You to parasar For This Useful Post:

     paulgill (01-18-2016)

  13. #7
    Banned
    Posts
    5,171
    Sex

    Quote Originally Posted by parasar View Post
    Were Ust-Ishim and the Denisova Neanderthal in this comparison?
    Also the Brahui and the Makrani, where do they fall?
    The dataset includes 2 Denisova, Altai Neand, and Vinija Neand. Brahui and Makrani along with 2000 other samples are in the dataset. Ust is not.

    The results I had posted above where prior to me identifying African Archaic matching allele frequencies at various loci. I have since identified and removed those SNPs while preserving sites where the 4 archaics have an allele freq >50% to the exclusion of Africans.

    This naturally biases the dataset to Archaic derived allele frequencies, but in my opinion that is ok, since I am trying to identify geneflow from archaic to modern. For obvious reasons, if there is a match between Archaics and moderns, this would most likely be due to geneflow, and not shared ancestral alleles basal to humans and Neanderthals.

    As expected the IBS percentages have drastically dropped across the board, as the frequencies at the loci that I extracted are very unique to Neanderthals only. By contrast, datasets containing moderns have matching allele frequencies at many loci

  14. The Following 2 Users Say Thank You to Kurd For This Useful Post:

     parasar (01-18-2016),  paulgill (01-18-2016)

  15. #8
    Gold Class Member
    Posts
    2,718
    Sex

    Quote Originally Posted by Kurd View Post
    The results I had posted above where prior to me identifying African Archaic matching allele frequencies at various loci. I have since identified and removed those SNPs while preserving sites where the 4 archaics have an allele freq >50% to the exclusion of Africans.
    Unless I misunderstand you, you seem to be saying that you threw out data that did not fit your preconceived model, which presumes that Africans have no archaic admixture. But research indicates that they did indeed.

    Genetic Evidence for Archaic Admixture in Africa by Hammer et al.
    ---
    Extensive simulation results reject the null model of no admixture and allow us to infer that contemporary African populations contain a small proportion of genetic material (≈2%) that introgressed ≈35 kya from an archaic population that split from the ancestors of anatomically modern humans ≈700 kya.
    ---

    Evolutionary History and Adaptation from High-Coverage Whole-Genome Sequences of Diverse African Hunter-Gatherers by Lachance et al.
    ---
    In all three African hunter-gatherer samples, we found evidence of introgression from at least one archaic population. Strikingly, the median TMRCA for putatively introgressed haplotypes in the hunter-gatherer samples is similar to the median TMRCA for introgressed haplotypes in Europeans (1.2–1.3 Mya versus 1.1–1.2 Mya, respectively; Figure 2A), suggesting that the archaic African population diverged from anatomically modern humans in the same time frame as Neanderthals (simulations suggest that relative time of split with archaic populations can be recovered via TMRCA; Figure 3C).
    ...
    In short, we find that low levels of introgression from an unknown archaic population or populations occurred in the three African hunter-gatherer samples examined, consistent with findings of archaic admixture in non-Africans (Reich et al., 2010).

    ---

  16. The Following User Says Thank You to lgmayka For This Useful Post:

     Tomenable (01-21-2016)

  17. #9
    Gold Class Member
    Posts
    7,801

    Quote Originally Posted by lgmayka View Post
    Unless I misunderstand you, you seem to be saying that you threw out data that did not fit your preconceived model, which presumes that Africans have no archaic admixture. But research indicates that they did indeed.

    Genetic Evidence for Archaic Admixture in Africa by Hammer et al.
    ---
    Extensive simulation results reject the null model of no admixture and allow us to infer that contemporary African populations contain a small proportion of genetic material (≈2%) that introgressed ≈35 kya from an archaic population that split from the ancestors of anatomically modern humans ≈700 kya.
    ---

    Evolutionary History and Adaptation from High-Coverage Whole-Genome Sequences of Diverse African Hunter-Gatherers by Lachance et al.
    ---
    In all three African hunter-gatherer samples, we found evidence of introgression from at least one archaic population. Strikingly, the median TMRCA for putatively introgressed haplotypes in the hunter-gatherer samples is similar to the median TMRCA for introgressed haplotypes in Europeans (1.2–1.3 Mya versus 1.1–1.2 Mya, respectively; Figure 2A), suggesting that the archaic African population diverged from anatomically modern humans in the same time frame as Neanderthals (simulations suggest that relative time of split with archaic populations can be recovered via TMRCA; Figure 3C).
    ...
    In short, we find that low levels of introgression from an unknown archaic population or populations occurred in the three African hunter-gatherer samples examined, consistent with findings of archaic admixture in non-Africans (Reich et al., 2010).

    ---
    As he is testing for archaic Denisovan introgression I think his approach is fine. Denisovan (and/or Neandertal) that is shared with Africans and but not with Eurasians, can be assumed to have mutated in Eurasians.

  18. The Following 4 Users Say Thank You to parasar For This Useful Post:

     Kurd (01-19-2016),  MonkeyDLuffy (01-18-2016),  paulgill (01-18-2016),  Tomenable (01-21-2016)

  19. #10
    Banned
    Posts
    5,171
    Sex

    Quote Originally Posted by lgmayka View Post
    Unless I misunderstand you, you seem to be saying that you threw out data that did not fit your preconceived model, which presumes that Africans have no archaic admixture. But research indicates that they did indeed.
    Igmayka,

    Yes, you misunderstood. Thanks for the links. This reminds me of a discussion we had on a different thread about a year ago, where I was trying to make sense of Dstats showing Mbuti much closer to chimp than Eurasians. We then contemplated this very idea of introgression of an archaic group into Mbuti to account for the unexpected stats. A very interesting discussion that I think we should pursue on a different thread.

    The SNPs that I removed that I was referring to, are sites where Africans and Neanderthal had similar allele frequencies to the exclusion of Eurasians. An archaic introgression into Africans at those sites, would have likely caused Africans to be different at those loci, since this happened much later than the Neanderthal African split. The sites I identified, I don't believe to be due to some Neanderthal African shared drift to the exclusion of Eurasians. I believe that they have simply
    changed in Eurasians for any number of reasons

  20. The Following 2 Users Say Thank You to Kurd For This Useful Post:

     paulgill (01-19-2016),  Tomenable (01-21-2016)

Page 1 of 4 123 ... LastLast

Similar Threads

  1. 1/2 Neanderthal 1/2 Denisovan DNA
    By euromutt in forum Ancient (aDNA)
    Replies: 2
    Last Post: 08-29-2018, 03:50 AM
  2. Neanderthal/Denisovan Haplogroups
    By Erik in forum Ancient (aDNA)
    Replies: 10
    Last Post: 08-27-2018, 04:08 PM
  3. Papuan's non-Denisovan ancestry
    By Kale in forum Autosomal (auDNA)
    Replies: 0
    Last Post: 07-20-2017, 05:21 PM
  4. From Denisovan DNA to future humanity
    By PLogan in forum Ancient (aDNA)
    Replies: 0
    Last Post: 03-18-2016, 03:34 PM
  5. Denisovan DNA
    By Edward J in forum Ancient (aDNA)
    Replies: 2
    Last Post: 11-19-2015, 11:36 AM

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •