Human DNA gets another ghost ancestor, right on schedule

TL;DR: A study published in Science and covered by The Debrief on July 31 used a new reference-free method called TRACE to scan genomes from the 1000 Genomes Project and found between 0.5 and 1.1 percent DNA from an unidentified archaic human, present in African and non-African populations alike, diverged from our line roughly 830,000 years ago. A second, older signal, dated to about 1.77 million years back, turned up riding inside Denisovan-derived DNA in Papua New Guinea, Vanuatu, and the Santa Cruz Islands. Neither ghost has a face, a name, or a bone to its name. I’ve shelved this exact headline before. More than once.

Six years on a returns desk teaches you a book’s spine before you ever crack it. You learn what a cover with a certain font and a certain blue means, you learn the shape of a plot from the back-cover copy alone, and you learn, eventually, that some stories in a genre aren’t new stories. They’re the same beat, redressed, run again because the beat works.

The ghost-ancestor beat works. The Debrief ran it this week under a headline about a “mysterious ghost human ancestor,” and by the numbers, the underlying paper earns it. TRACE, short for TRacking Archaic Contributions via ARG Estimation, reconstructs ancestral recombination graphs instead of comparing genomes against a known archaic reference. That matters because the population it’s detecting left no bones anyone has found. No skull, no femur, no reference genome to check against. Just the long branches and odd patterns the interbreeding left behind in the DNA of everyone alive now, an estimated 0.5 to 1.1 percent of it, spread across more than 70 percent of the accessible human genome, sitting in some spots people had written off as archaic-free “deserts.”

That’s the paper. Here’s the shelf it sits on.

I want to be fair to it before I file it. TRACE was tested against extensive simulations of known Neanderthal interbreeding before anyone pointed it at real data, and the team reports it came back accurate with a low false-positive rate. That’s the part a returns-desk skeptic actually respects. A tool that finds a ghost and can also tell you when there isn’t one.

The first time I remember this exact chapter running, the title was “Genetic evidence for archaic admixture in Africa,” a PNAS paper from 2011, Michael Hammer’s group, finding roughly 2 percent unidentified archaic DNA in some African genomes with no reference sample to pin it to. The coverage at the time used the same word this week’s coverage uses. Ghost. Then in 2020, Arun Durvasula and Sriram Sankararaman published “Recovering signals of ghost archaic introgression in African populations” in Science Advances, and that title is the beat’s name tag, not an echo of the beat. Different statistical machinery, same premise: an African population’s genome holds a signature from something we can’t put a face to, so the paper goes looking for the shape of the absence instead of the thing itself.

What’s actually different this time isn’t the ghost. It’s where TRACE found it. The 2011 and 2020 work stayed inside Africa, because that’s where the archaic-free reference comparison broke down hardest and the signal was strongest to detect with older tools. This paper finds a similar-sized ghost signature in Africans and non-Africans both, and argues the shared portions outside Africa mean the interbreeding happened before the out-of-Africa dispersal, not after, and not locally. That’s a genuinely different claim than the last two times this beat ran. It moves the ghost from a regional footnote to a chapter everyone’s genome shares.

The other new piece is the older ghost, the one dated to 1.77 million years, detected inside Denisovan-derived segments in Oceanian populations. Earlier papers on a “third archaic lineage” in Island Southeast Asia and Melanesia gestured at something like this without dating it this precisely or tying it this directly to Denisovan ancestry as the carrier. If it holds up, the old beat does not cover it. What you have instead is a ghost inherited secondhand, through another ghost.

I don’t have a verdict on whether TRACE will still be cited in five years, and neither does anyone quoted in this piece. Method papers in this space have a way of getting superseded by the next reference-free trick, and “ghost ancestor” headlines have a way of running again in eighteen months regardless of which method wins. What I can tell you is which of the two claims in this one is worth remembering past the headline, and the percentage is not it. The geography is.

Sources

• The Debrief: Scientists Discover Evidence of a Mysterious ‘Ghost’ Human Ancestor Hidden in Our DNA