henhao
Sign in
Evolutionary BiologyScience
Peer-reviewed journal article

A global map for introgressed structural variation and selection in humans

PingHsun HsiehiD, Natthapon SoisangwaniD, David GordoniD, Athef Javidh, William T. HarveyiD, David PorubskỳiD, Kendra HoekzemaiD, Carl A. Baker, Katherine M. MunsoniD, Christopher KinipiiD, Matthew LeavesleyiD, Nicolas BrucatoiD, Murray P. CoxiD, François‐Xavier RicautiD, Irene Gallego RomeroiD, Evan E. EichleriD

Abstract

Genetic introgression from Neanderthals and Denisovan has shaped modern human genomes; however, introgressed structural variants (SVs [≥]50 base pairs) remain challenging to discover. We integrated high-quality phased assemblies from four new Papua New Guinea (PNG) genomes with 94 published assemblies of diverse ancestry to infer an archaic introgressed SV map. Introgressed SVs are overall enriched in genes (44%, n=1,592), including critical genomic disorder regions, and most abundant in PNG. We identify 11 centromeres likely derived from archaic hominins, adding unexplored diversity to centromere genomics. Pangenome genotyping across 1,363 samples reveals 16 candidate adaptive SVs, many associated with immune-related genes and their expression, in the PNG. We hypothesize that archaic SV introgression contributed to reproductive success, underscoring introgression as a significant force in human adaptive evolution.

PDFDOI
Full-text digest ↓

02 /Specification & Classification

Click any tag to browse matching papers
Citations
5
FWCI
83.75
Field percentile
top 1%
OpenAlex Topics

In-depth summary

✦ Multi-agent verified full-text synthesis

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (24 Jun 2025) · CC BY 4.0

Long-read phased assemblies from two Papuans plus 47 HPRC genomes map archaic introgressed structural variants, revealing 11 candidate archaic centromeres and 16 candidate adaptive SVs.

Introgressed structural variants (SVs) are hard to detect because ancient DNA is highly fragmented and short reads miss over 70% of SVs, notably in repeats, segmental duplications and centromeres. Adding Papuan genomes, missing from 1000 Genomes-derived HPRC panels yet richest in archaic ancestry, to phased long-read assemblies enables an assembly-based SV introgression map beyond earlier short-read-centric studies.

Key results

  1. ShownPapuan genomes carry about twofold more archaic-derived DNA than other non-African HPRCr1 samples, and thousands of insertions and deletions map onto introgressed Neanderthal- or Denisovan-like haplotypes.

    445 Mbp introgressed sequence across four PNG haplotypes (3.82% of PNG15, 4.37% of PNG16); 1,419 deletions and 2,137 insertions in 1,807 segments; mean archaic fraction 4.10% PNG vs 1.34% AMR (n=16), 2.13% EAS (n=5), 1.19% EUR, 1.72% SAS; about 335 Mbp nonredundant introgressed DNA (11.2% of genome) across 25 individuals. · A global map of genomic introgression from…

  2. SuggestedPangenome genotyping in short-read cohorts identifies 16 highly differentiated SVs in 13 selected Papuan regions, including a Denisovan-like 61 bp intronic insertion in CEACAM18 and a 395 kbp insertion spanning NPIPB16.

    71 PNG, 703 African and 585 East Asian genomes genotyped; 16 SVs at median PBS_SV 0.79 (p<0.002; top 1% cutoff 0.39). 61 bp insertion: 68% in PNG vs 13% elsewhere, s = 0.008, p = 0.0088. 395 kbp insertion: 62.6% in PNG, absent elsewhere, s = 0.005, p = 0.023, strong selection from about 5,800 years ago. · Adaptive introgression in PNG genomes

  3. SuggestedAn alignment-free archaic k-mer scan flags 11 of 55 complete Papuan centromeres as likely archaic-derived; chromosome 4 and 22 candidates show diverged alpha-satellite higher-order repeat (HOR) structure and deep divergence times.

    5 (PNG15) and 6 (PNG16) of 55 complete centromeres exceed archaic k-mer thresholds; chr4 HOR identity 99.8% modern-modern vs 98.8% archaic-modern (one-sided MWU p < 2.2×10^-16); BEAST divergence 489,447–507,131 years (chr4) and 580,661 years (chr22); HOR mutation rates 2.94×10^-8 (chr4) and 7.46×10^-8 (chr22) per generation. · Potential centromere introgression from archaic…

  4. ShownIntrogressed SVs are modestly enriched in and near genes in every continental group, unlike PNG-specific SVs, which are depleted in genes; enriched GO terms include neural guidance and calcium transport.

    44% of introgressed SVs (n=1,592) overlap genes; in PNG, 710 introgressed SVs lie within 1 kbp of genes, odds ratio 1.09 (95% CI 1.04–1.15, one-sided Fisher p=0.0056); negative chemotaxis 4.61-fold (Bonferroni p=2.2×10^-2), calcium ion transmembrane transport 2.54-fold (p=6.6×10^-3); PNG-specific SVs gene-depleted (p<0.001, 1,000 permutations). · Genomic variation in the Papuan assemblies; A global map of genomic introgression from…

  5. ShownFour near-complete phased Papuan haplotypes outperform GRCh38 and most HPRCr1 assemblies in contiguity and reveal unique variation, plus a rare 0.68 Mbp deletion, also seen in three HPRCr1 haplotypes, that removes much of LCR22A at the 22q11.2 deletion syndrome locus.

    Verkko haplotypes of 2.89 to 3.01 billion bp, scaffolded N50 111–136 Mbp, mean QV 44.5, >99.3% accurately assembled; 12% of small variants and 5.6% of SVs PNG-specific (East Asian pairs 4.7–4.9%); 22q11.2 deletion breakpoints lie in FAM230D/FAM230F SDs sharing >98.8% identity, structure matching chimpanzee. · High-quality, haplotype-phased Papuan assemblies; Genomic variation in the Papuan assemblies; DISCUSSION

Methods, data and reproducibility

Two Papuan lymphoblastoid lines were sequenced (HiFi, ultra-long ONT, Hi-C, Illumina) and assembled with Verkko; PAV-led SV calls were projected onto Sprime- and HMM-inferred introgressed segments, genotyped with PanGenie in short-read cohorts for PBS/fD selection scans and CLUES, while an archaic k-mer scan and BEAST dating probed centromeres.

  • New phased long-read assemblies of PNG15 (Western Lowland PNG female) and PNG16 (Solomon Islands male) lymphoblastoid cell lines · human · n = two individuals; four haplotype assemblies
  • Human Pangenome Reference Consortium Release 1 (HPRCr1) haplotype assemblies and PAV call set · human · n = 47 individuals
  • Published high-coverage (>20×) Papua New Guinea short-read genomes genotyped via pangenome graph · human · n = 71
  • High-coverage 1000 Genomes Project short-read samples used as outgroup and sister group · human · n = 703 African; 585 East Asian
  • High-coverage archaic hominin genomes (Vindija Neanderthal and Denisovan among them) · Neanderthal and Denisovan (archaic hominins) · n = four genomes
  • Reference assemblies GRCh38 and T2T-CHM13v1.1 for variant calling and alignment · human
  • High-quality chimpanzee assembly used to infer ancestral state of the 22q11.2 deletion · chimpanzee
  • GTEx portal eQTL associations for SNVs linked to the CEACAM18 insertion · human (GTEx)
Methods & tools
Verkko v1.4.1, PAV v1.1.2, SVIM-ASM v1.0.3, Sniffles2 v2.2, PBSV v.2.9.0, Flagger, NucFreq, Merqury, compleasm (assembly QC), Sprime, HMM-based archaic introgression caller, PanGenie, msprime coalescent simulations, CLUES and pairwise coalescent decoding, BEAST, StainedGlass, Archaic-specific k-mer scan (alignment-free)
Reagents & environment
Verkko v1.4.1; PAV v1.1.2; SVIM-ASM v1.0.3; Sniffles2 v2.2; PBSV v.2.9.0; RefSeq release 109 gene annotation; GRCh38 and T2T-CHM13v1.1 references.
Reproducibility gaps
European Genome-Phenome Archive accession given only as placeholder EGASXXXXXX pending approval; no code repository or URLs provided; detailed methods deferred to Supplementary Information.
Limitations (6)
  • Only 49 individuals (98 haplotype assemblies) were analyzed, covering a small share of global human genetic diversity. · stated by the authors
  • PNG15 and PNG16 show more Neanderthal (1.96%, 2.22%) than Denisovan (1.25%, 1.55%) ancestry, contrary to expectation; authors cite mismatch between the sequenced Denisovan and the admixing population, and sampling of lowlanders only. · stated by the authors
  • Inversions had <15% cross-caller support and over 90% were sample- or cohort-specific, so introgression and selection analyses largely exclude them and center on insertions and deletions. · stated by the authors
  • Short-read pangenome genotyping performs poorly at repetitive loci (VNTRs, subtelomeric and centromeric regions) and deletion concordance drops with size; after filtering, 199,917 of 613,375 SVs remained. · stated by the authors
  • Alpha-satellite HOR mutation rates are approximations assuming a 400-700 kya divergence and do not model rapid HOR turnover or structural change. · stated by the authors
  • Functional roles of candidate adaptive SVs (CEACAM18 intron, PTPRH upstream, NPIPB16) rest on eQTL, H3K27ac and selection statistics without functional assays; the proposed 22q11.2 protective effect is untested. · noted in this summary

Who should read:Read this if you study archaic introgression, structural variation in underrepresented populations, or centromere evolution and want assembly-based SV maps with selection scans in Papuan genomes.

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISYan '21Suo '26Hu '15Vespasiani '20Skov '18Gouy '19
72% semantic matchEvolutionary Biology
S. M. Yan, R. M. Sherman, D. J. Taylor+4 more·bioRxiv·Jan 26, 2021