In 1980, a monk praying in a limestone cave on the north-eastern edge of the Tibetan Plateau found half a human jawbone. The cave, Baishiya Karst, sits at 3,280 metres above sea level near the town of Xiahe, and the bone passed through a local monastery before it reached scientists at Lanzhou University. It had no DNA left in it. In 2019, a team extracted ancient proteins from the dentine of its two remaining molars and compared them to the known hominins. The jaw belonged to a Denisovan, and the carbonate crust that had grown over it put its age at 160,000 years or more.1
I find that date hard to hold in my head. Modern humans did not settle the plateau until tens of thousands of years later. Long before us, a different kind of human was already living in thin air at the roof of the world.
That detail matters because of what Tibetans carry today. At 4,000 metres, the air holds about 60 percent of the oxygen it does at sea level. Lowlanders who move there respond the way physiology textbooks say they should: their bodies make more red blood cells, haemoglobin rises, and the blood thickens. Over years that can tip into chronic mountain sickness, with headaches, clotting, and heart strain. Tibetans mostly don't do this. Their haemoglobin stays close to normal, and in 2010 two independent studies traced the reason to a gene called EPAS1, which encodes a master switch in the body's response to low oxygen. Tibetans carrying the high-altitude version had measurably lower haemoglobin than their neighbours without it.2
Four years later, Emilia Huerta-Sánchez and Rasmus Nielsen's group looked more closely at the stretch of DNA around EPAS1. The Tibetan version was so different from every other human version that it could not have arisen by ordinary mutation in the time available. It matched the Denisovan genome. The haplotype was present on 69 of the 80 Tibetan chromosomes they sequenced, turned up in two individuals among thousands of Han Chinese, and appeared nowhere else in the world. The only reasonable explanation was that Denisovans had passed it to the ancestors of Tibetans.3 A 2021 follow-up estimated that the segment entered the population around 48,700 years ago and sat quietly for tens of thousands of years before selection started favouring it, around 9,000 years ago, roughly when people began living on the plateau year-round.4
So the people who breathe most easily at altitude do it with a gene from a species that was breathing that same air 160,000 years ago. That is the clearest case we have, but it is not the only one. Almost everyone outside Africa carries a slice of DNA from humans who are no longer here, and some of it is still doing things.
A finger bone in Siberia
Denisovans were discovered backwards. There was no skull, no skeleton, no stone tools that could be tied to them. In 2008, Russian archaeologists excavating Denisova Cave in the Altai Mountains found a fragment of a finger bone, the last segment of a little finger, and sent it to Svante Pääbo's lab at the Max Planck Institute for Evolutionary Anthropology in Leipzig, which had spent a decade learning to pull DNA out of Neanderthal bones. Everyone assumed it was a Neanderthal or an early modern human.
It was neither. The mitochondrial DNA, published in March 2010, belonged to a lineage that had split from both of us roughly a million years ago.5 By December the lab had a draft of the whole genome. It showed the bone's owner belonged to a sister group of Neanderthals, and it showed something nobody had predicted: present-day people in Melanesia carry 4 to 6 percent of their DNA from this group. The authors named them Denisovans, after the cave.6 Two years later, a new method for building sequencing libraries squeezed a 30-fold-coverage genome out of the same few milligrams of bone, as good as a genome from a living person, and revealed that the finger had belonged to a girl.7
For years, that finger bone, a few teeth and the Xiahe jaw were all the physical evidence of Denisovans that existed. An entire human population known mainly as a pattern in other people's DNA. Pääbo received the Nobel Prize in 2022 for this body of work.8 Only in 2025 did Denisovans get a face. A nearly complete skull from Harbin in north-eastern China, at least 146,000 years old and previously given its own species name, yielded 95 ancient proteins, three of which carried Denisovan-specific changes. The skull clustered with the Denisova Cave girl.9 We now know what they looked like: a broad face, a heavy brow, a brain as large as ours.
The genetic trail is also messier than one cave suggests. Papuans carry DNA from two deeply separated Denisovan lineages that had been apart for more than 350,000 years, and East Asians carry a third. People mixed with Denisovans in several places, at several times, probably over tens of thousands of years.10
How much, and when
Neanderthals were never so mysterious. We have had their bones since 1856. What we didn't have until 2010 was their genome, and when Pääbo's team published a draft that year, assembled from three bones found in a Croatian cave, the headline finding was that Neanderthals shared more variants with living Europeans and Asians than with living Africans. The simplest explanation was interbreeding, after modern humans left Africa and before they spread across Eurasia. The estimate was that 1 to 4 percent of the genome of every non-African came from Neanderthals.11 A better Neanderthal genome in 2017, from a woman who lived in Vindija Cave about 50,000 years ago, sharpened that and let researchers recover 10 to 20 percent more Neanderthal DNA in modern people than before.12 The usual figure now is about 2 percent.
Two percent sounds small until you notice that it is a different 2 percent in each of us. My Neanderthal fragments are not yours. Pooled across living people, a large share of the Neanderthal genome is still in circulation, just scattered.
The timing has also come into focus. A 2024 study that catalogued Neanderthal segments in more than 300 ancient and modern genomes spanning the last 50,000 years found that almost all of the mixing happened in a single extended period between about 50,500 and 43,500 years ago.13 That is the window in which modern humans were pushing into Europe and Asia. It wasn't an occasional encounter. For several thousand years, these were neighbours.
What it does
For a while the fashionable view was that archaic DNA was mostly inert, a fossil inside us. It isn't. Some of the clearest signals are in immunity, which makes sense. Neanderthals had lived in Eurasia for hundreds of thousands of years and had adapted to its pathogens. The modern humans arriving from Africa hadn't. Borrowing a tuned immune gene was faster than evolving one.
A cluster of three Toll-like receptor genes, which sit on the surface of immune cells and recognise bacteria and fungi, carries three distinct archaic versions in people today. Two match Neanderthals and one matches Denisovans. Their frequency is far too high to be chance, and they are associated with increased resistance to infection and, as the bill for that, increased allergic disease.14 Survive the pathogen, pay with hay fever.
Beyond immunity, the picture is real but modest. A 2016 study matched Neanderthal variants against electronic health records from about 28,000 adults and found associations with depression, with skin lesions caused by sun exposure, and with blood clotting.15 A UK Biobank analysis of 112,000 people the following year found Neanderthal DNA influencing skin tone, hair colour, height, sleeping patterns, mood and smoking. The skin result is the one I like most, because it cuts against the cartoon version. Neanderthal variants contribute to both lighter and darker skin, which suggests Neanderthals themselves varied, the way we do.16 Regions with high Neanderthal ancestry are enriched for keratin genes, which build skin and hair.17 None of these effects is large. Nobody is a morning person because of their Neanderthal ancestry alone. And a more careful 2022 analysis found no excess of Neanderthal variants in psychiatric or neurological disease, only in behavioural traits like pain sensitivity, sleep timing and smoking.22 The headlines about Neanderthal genes causing depression ran well ahead of the data.
The pandemic
The example that made this real to most people came in 2020. The first genome-wide studies of severe COVID-19 found that the strongest risk signal sat in a cluster of genes on chromosome 3. Hugo Zeberg and Pääbo looked at the region and recognised its shape. The risk variants travelled together on a long segment of about 50,000 letters, the signature of DNA that entered our species recently from an outside source, and the alleles matched the Vindija Neanderthal genome almost perfectly.18
The distribution was striking. About half of South Asians carry the segment, and roughly 16 percent of Europeans. In Bangladesh, 63 percent of people carry at least one copy and 13 percent carry two. It is nearly absent in East Asia and in Africa. Each copy roughly doubles the risk of needing intensive care after infection.1819 During the first year of the pandemic, people of Bangladeshi origin in the UK were dying of COVID-19 at about twice the rate of the general population, and while poverty, housing and occupation surely did most of that work, some part of it was a piece of DNA from a woman who lived in Croatia 50,000 years ago.
Then the same two authors found the opposite. A region on chromosome 12, containing the OAS genes that help cells destroy viral RNA, carries a second Neanderthal haplotype. This one cuts the risk of needing intensive care by about 22 percent per copy. It is found at 25 to 30 percent frequency across most of Eurasia, and ancient genomes show its frequency rising between 20,000 and 10,000 years ago and again in the last thousand years, as if earlier epidemics had already been selecting for it.19 The same vanished neighbours gave us one variant that made the pandemic worse and another that made it better. There is no moral to that. It is just what inheritance looks like.
The deserts
The parts of the genome where archaic DNA is missing are as telling as the parts where it survives. When Sriram Sankararaman and David Reich mapped Neanderthal ancestry across 1,004 genomes in 2014, they found long stretches with almost none of it. These deserts are rich in genes, which means selection has been actively scrubbing Neanderthal versions out. Two patterns stood out. Genes that are expressed mostly in the testes are especially depleted. And the X chromosome carries about five times less Neanderthal ancestry than the rest of the genome.17 In other animals, that is the classic fingerprint of hybrids with reduced male fertility. The same pattern holds for Denisovan DNA.20 The sons of these unions, some of them, may have struggled to have children of their own.
There is a quieter reason the Neanderthal share has shrunk. Neanderthal populations were small, far smaller than ours, and small populations accumulate mildly harmful mutations that selection is too weak to remove. Those mutations were harmless enough among Neanderthals. In the much larger human population they met, selection could finally see them, and it has been slowly weeding them out ever since.21 Most of that pruning happened fast, within a few thousand years of the mixing.13 What remains today is, in a sense, the part that passed the test.
A mixture, all the way down
It is tempting to tell this as a story about us and them, in which modern humans picked up a few useful spare parts from species that then went extinct. I don't think that framing survives contact with the data. Neanderthals and Denisovans didn't entirely go anywhere. Their lineages ended as separate populations, but their DNA is sitting in intensive care units and on the Tibetan Plateau right now, affecting who breathes easily and who ends up on a ventilator. The woman from Vindija Cave has, by any reasonable count, hundreds of millions of living descendants.
The deeper lesson is that there never was a single, pure modern human lineage to be contaminated. Modern humans mixed with Neanderthals for thousands of years, with at least three Denisovan groups across Asia, and almost certainly with other populations in Africa whose genomes we haven't yet read. Being human has always meant being a mixture. The two percent that isn't yours turns out to be a reasonable description of all of us.