Report

The phenotypic legacy of admixture between modern humans and Neandertals

See allHide authors and affiliations

Science  12 Feb 2016:
Vol. 351, Issue 6274, pp. 737-741
DOI: 10.1126/science.aad2149
  • Fig. 1 Analysis of EHRs reveals clinical effects of Neandertal alleles in modern humans.

    (A) Thousands of Neandertal alleles were identified in ~28,000 individuals of European ancestry across the eMERGE Network. We derived phenotypes for each individual from data in their EHRs. (B) To test Neandertal alleles in aggregate for phenotype associations, we computed the genetic similarity of all pairs of individuals over 1495 genotyped Neandertal loci and their phenotypic similarity over 46 EHR-derived traits. (C) We estimated the overall variance in risk explained by Neandertal alleles using mixed linear models in GCTA (15) and found that Neandertal alleles explain significant variance in several traits (Table 1). (D) To test individual Neandertal alleles for trait associations, we performed a discovery meta-analysis across eMERGE E1 sites with sufficient data. We then ran a replication meta-analysis over the independent eMERGE E2 cohort. This approach identified and replicated several associations (Table 2). The example forest plot illustrates the association of Neandertal SNP rs3917862 with hypercoagulable state in each site with >=20 cases for the separate discovery and replication analyses. (E) rs3917862 is located in an intron of P-selectin (SELP), a gene that mediates leukocyte action at injuries in the early stages of inflammation. The Neandertal allele is significantly associated (linear regression, P = 0.005) with increased expression of SELP in tibial artery data from GTEx.

  • Fig. 2 Neandertal SNPs associate with different phenotypes than matched non-Neandertal SNPs.

    Each bar gives the difference between the number of replicated Neandertal SNP associations with a phenotype group (at a relaxed discovery threshold of P < 0.001) and the number expected from a PheWAS over five sets of non-Neandertal sites matched to the allele frequency of tested Neandertal SNPs. The phenotype distributions were significantly different (chi-squared test, P = 0.017), with more Neandertal SNPs associated with neurological and psychiatric phenotypes than expected and fewer digestive phenotypes. The enrichment and depletion were consistent across all five matched non-Neandertal sets (* indicates P < 0.05 for all five comparisons; binomial test) (11).

  • Table 1 Neandertal alleles explain risk for human clinical traits.

    The eight traits for which Neandertal alleles explained a nominally significant proportion of variance in risk in both the E1 discovery and E2 replication analyses are listed (GCTA, P < 0.1). The depression association remained significant after controlling the false discovery rate at 5%. The Neandertal associations with actinic keratosis, mood disorders, and depression were also maintained in a two-GRM model that considered the risk explained by non-Neandertal variants. Phenotypes are sorted by their E2 P value.

    PhenotypeDiscovery (E1)Replication (E2)Replication
    (E2; two-GRM)
    Risk explainedPRisk explainedPRisk explainedP
    Actinic keratosis0.64%0.0663.37%0.00592.49%0.036
    Mood disorders1.11%0.00910.75%0.0180.68%0.029
    Depression2.03%0.00231.15%0.0201.06%0.031
    Obesity0.59%0.0481.23%0.0300.39%0.27
    Seborrheic keratosis0.77%0.0380.61%0.0450.41%0.13
    Overweight0.60%0.0370.53%0.0520.23%0.24
    Acute upper respiratory infections0.70%0.0430.56%0.0620.34%0.18
    Coronary atherosclerosis0.68%0.040.42%0.0980.34%0.15
  • Table 2 Individual Neandertal SNPs with significant replicating phenotype associations.

    Four locus-wise Bonferroni significant Neandertal SNP–phenotype associations replicated (with a fixed effect P < 0.05 and consistent direction of effect). Nominally significant replicating results can be found in table S3 and in the PheWAS Catalog (https://phewas.mc.vanderbilt.edu/neanderthal). Chr, chromosome.

    DiscoveryReplication
    PhenotypeChr:position
    (hg19)
    SNPFlanking
    gene(s)
    Odds
    ratio
    POdds
    ratio
    P
    Hypercoagulable
    state
    1:169593113rs3917862SELP3.329.9 × 10−73.005.0 × 10–10
    Protein-calorie
    malnutrition
    1:234099819rs12049593SLC35F31.772.0 × 10−61.635.5 × 10−5
    Symptoms involving
    urinary system
    11:3867350rs11030043RHOG, STIM11.767.4 × 10−61.654.3 × 10−2
    Tobacco use
    disorder
    3:10962315rs901033SLC6A112.191.7 × 10−51.757.9 × 10−4

Supplementary Materials

  • The phenotypic legacy of admixture between modern humans and Neandertals

    Corinne N. Simonti, Benjamin Vernot, Lisa Bastarache, Erwin Bottinger, David S. Carrell, Rex L. Chisholm, David R. Crosslin, Scott J. Hebbring, Gail P. Jarvik, Iftikhar J. Kullo, Rongling Li, Jyotishman Pathak, Marylyn D. Ritchie, Dan M. Roden, Shefali S. Verma, Gerard Tromp, Jeffrey D. Prato, William S. Bush, Joshua M. Akey, Joshua C. Denny, John A. Capra

    Materials/Methods, Supplementary Text, Tables, Figures, and/or References

    Download Supplement
    • Materials and Methods
    • Supplementary Text
    • Figs. S1 to S4
    • Tables S1 and S3 to S5
    • Caption for Table S2
    • References (26–51)
    Table S2
    GREAT analysis of Neanderthal SNPs with highest/lowest BLUPs from actinic keratosis and depression GCTA tests.

Navigate This Article