This Halloween, you can skip the ghostly sheet. Every human already carries a ghost, several of them, in fact, stitched into the long strands of DNA we inherited from our parents. It was passed down from long-dead populations of our ancestors who vanished so completely that no one has ever found their bones.
The ancestors of modern humans, known as hominins, interbred with 2 extinct relatives, known as archaic humans. Most people outside Africa carry roughly 1% to 2% Neanderthal DNA. Their DNA began mixing with ours more than 50,000 years ago and continued for roughly 7,000 years, until they disappeared. People in Asia and Oceania also carry between 0.1% and 5% DNA from another population, known as the Denisovans.
Scientists have pieced together this ancestry by comparing living human genomes with DNA recovered from fossil hominins. However, researchers have read the full DNA code of only 6 of these individuals: 4 Neanderthals and 2 Denisovans, all from Eurasia. Other hominin groups lived alongside them, including Homo heidelbergensis in Africa and Homo erectus in Asia, but nobody has tested whether we carry their DNA. Ancient DNA degrades quickly outside of permanently frozen areas, so the odds of reading the code of someone who lived a million years ago in a warm climate are near zero.
That lack of ancient DNA creates a problem for scientists reconstructing human evolution. If another population interbred with our ancestors and then vanished, how would we even know it existed? Scientists call a group whose only surviving trace is the DNA it left inside a living organism a ghost lineage.
Scientists at Berkeley and Johns Hopkins recently used the genomes of modern humans to test whether ghost lineages are hiding in our DNA. Human DNA comes in 23 long strands, called chromosomes. Each short stretch along a strand has its own history, passed down through one particular chain of ancestors. Because our chromosomes get reshuffled every generation, adjacent stretches may have travelled completely different routes. A single chromosome therefore carries thousands of separate family trees side by side, which geneticists compile into an ancestral recombination graph or ARG.
Timing is what makes an ARG useful for reconstructing ancestry. Follow any stretch of DNA backwards, and you’ll eventually reach the single ancestor who first carried that stretch. For most humans, that ancestor lived fairly recently. But stretches inherited from extinct relatives reach much further back because those lineages parted ways hundreds of thousands of years earlier. Therefore, the team wrote a computer program to scan these family trees for branches reaching unexpectedly far back, and named it Tracking Archaic Contributions via ARG Estimation, or TRACE.
First, the researchers tested TRACE on simulated genomes, where it correctly identified the stretches they deemed archaic. Then they turned it on 503 human genomes from a public catalogue of donated DNA called the 1000 Genomes Project, covering 5 human populations: British, Han Chinese, Indian Telugu, Yoruba from Nigeria, and Luhya from Kenya. Earlier researchers found archaic DNA by matching it against a sequenced archaic genome. TRACE found the same stretches without one, suggesting it could also detect relatives without recorded fossils.
The team further identified some segments in these genomes that traced back to a ghost population of hominins that separated from our lineage roughly 830,000 years ago, around the same time the Neanderthals and Denisovans split off. Every population the team examined carried 0.5% to 1.1% of their DNA from these ghosts, suggesting that this group interbred with modern humans in Africa before they migrated into Eurasia.
They also discovered a second, far older lineage in genomes from Oceania, where Denisovan DNA is most common. This super-archaic population split off around 1.8 million years ago. Its DNA always sits inside stretches inherited from Denisovans, suggesting this ghost population interbred with Denisovans, who later interbred with our ancestors, carrying the older DNA with it.
The researchers concluded that these 2 ghost lineages account for about 2% of the modern human genome, suggesting that our family tree repeatedly separated and remixed. The identity of both ghost populations remains uncertain, although the first split from us around the same time as Homo heidelbergensis, and the second is old enough to be Homo erectus. Until someone puts a face to them, the ghosts haunting our DNA will remain anonymous. But unlike the ones knocking on your door this October, they are entirely real.
