The lost genome inside every living person

TL;DR

A study published 30 July 2026 in Science describes TRACE (TRacking Archaic Contributions via ARG Estimation), a computational method that detects archaic genetic contributions using present-day human genomes alone — no fossil DNA required.
Applied to 503 contemporary genomes from the 1000 Genomes Project, TRACE identified a previously unsequenced "ghost" ancestry signal in every population examined, generally accounting for about 0.5–1.1% of an individual's genome. This lineage diverged from modern humans roughly 830,000 years ago and mixed back in more than 50,000 years ago, before the major out-of-Africa migration.
A second, deeper "super-archaic" signal — diverging roughly 2 million years ago — appears to have entered the Denisovan lineage before reaching some modern people (particularly in Oceania) through Denisovan ancestry.
The ghost ancestry appears concentrated in genomic regions linked to immune defences and metabolism, and — surprisingly — in regions previously thought to reject all archaic DNA, including near FOXP2, a gene involved in speech and language.
The result is a methodological advance, not direct evidence of a newly named human species. No fossil has been found and no unknown ancient individual has been sequenced.

The missing archive

A human fossil can be silent for a million years. A genome is less cooperative, but it leaves clues.

The usual route to finding an extinct human relative is brutally constrained. A bone or tooth must survive. It must be found — which usually means it must erode out of sediment at the right moment, in a place where someone is looking. Its DNA must survive too, something warmth, water, and time are particularly good at preventing.

That helps explain why Neanderthal and Denisovan genomes are disproportionately informative: their remains were recovered from caves in Europe and Siberia, where cool, dry conditions preserved DNA for tens of thousands of years. Much of Africa, where modern human evolution unfolded across hundreds of thousands of years, does not offer those same conditions. Heat and humidity degrade DNA rapidly. The absence of ancient DNA from Africa has never meant the absence of ancient populations. It has meant the archive was inaccessible.

A team led by researchers at the University of California, Berkeley — including graduate student Yulin Zhang, postdoctoral researcher Arjun Biddanda, and principal investigator Priya Moorjani — has now demonstrated a way around that constraint. Their method, TRACE, published 30 July in Science, does not require a fossil. It reads the genealogical record embedded in the genomes of living people.

How TRACE works

TRACE stands for TRacking Archaic Contributions via ARG Estimation. The "ARG" — ancestral recombination graph — is the key concept.

Every human genome is a mosaic of DNA segments inherited from different ancestors. When a child is conceived, maternal and paternal chromosomes exchange segments through recombination, shuffling the genetic deck. Over generations, these recombination events create a branching, interweaving genealogical structure — the ARG — that records which segments of DNA were inherited from which ancestors, and when.

TRACE reconstructs these ARGs across hundreds of contemporary genomes. It then looks for segments whose genealogical branches are unusually long and divergent — DNA that appears to have separated from the modern human lineage far earlier than expected and then re-entered it through interbreeding. These are the signatures of archaic introgression: genetic material contributed by populations that split from our ancestors hundreds of thousands of years ago and later mixed back in.

Previous methods for detecting archaic ancestry required a reference genome from a known archaic individual — a sequenced Neanderthal or Denisovan — to serve as a comparison. TRACE needs only the genomes of living people. As Moorjani told Berkeley News: "We no longer have to wait for extraordinary fossil preservation to learn about extinct human populations. Instead, the genomes of living people contain traces of these ancient ancestors, and genealogical methods allow us to recover some of that hidden history."

Validation: finding what we already know

The team first tested TRACE on what it should find: known Neanderthal and Denisovan contributions.

Applied to genomes from the 1000 Genomes Project — encompassing hundreds of individuals representing populations across Africa, Europe, Asia, and elsewhere — TRACE correctly identified Neanderthal ancestry at the expected levels (roughly 1–2% in non-African populations). Some of the oldest genetic regions identified by the method matched DNA from a sequenced Neanderthal genome.

When the team focused on genomes from Asia and Oceania, where some populations carry substantial Denisovan ancestry, TRACE successfully identified known Denisovan DNA segments.

The researchers also ran extensive computer simulations modeling ancient human populations and known episodes of Neanderthal interbreeding. The method consistently demonstrated high accuracy with very low false-positive rates.

Having validated the approach against known signals, the team then asked: what else is there?

Finding 1: A ghost lineage in everyone

TRACE identified a previously unsequenced archaic signal in all populations examined — African, European, Asian, and others. The estimate varies by population but is roughly 0.5–1.1% of an individual's genome.

The study infers that this population split from the lineage leading to modern humans roughly 830,000 years ago — long before Homo sapiens emerged as a distinct species roughly 300,000 years ago. The admixture event — when this population mixed back into the modern human lineage — occurred more than 50,000 years ago, before the major migration out of Africa.

This is significant because it challenges a previous assumption. Earlier studies had suggested that ghost ancestry might be present only in specific African populations. Zhang and colleagues showed that it is present in all modern humans — meaning the interbreeding event occurred before our ancestors dispersed across the globe.

"Previous publications suggested that there might be ghost ancestry — ancestry from unknown archaic lineages in modern humans — but they hadn't concluded whether this unknown ancestry is present only in Africans or not, and when this introgression event happened," Zhang told Berkeley News. "We were actually able to find and map genomic locations in modern humans that are from this ghost lineage and show that this ghost ancestry is in all modern humans, not only in Africans."

Finding 2: A super-archaic lineage through Denisovans

A second, older signal — diverging roughly 2 million years ago — appears to have entered the Denisovan lineage first, with fragments later reaching modern humans through Denisovan ancestry. This "super-archaic" contribution is detectable primarily in Oceanian populations, who carry the highest levels of Denisovan ancestry.

This finding adds a layer to an already complex picture. It suggests that Denisovans themselves were not a genetically isolated population but had interbred with an even older, more divergent hominin lineage — and that some of that older lineage's DNA was then passed to modern humans through Denisovan intermediaries.

Where the ghost DNA lives — and what it might do

One of the study's most intriguing findings concerns where the ghost ancestry appears in the genome.

Many of the archaic DNA segments were concentrated in genomic regions connected with immune defences and metabolism, according to Moorjani. This is consistent with a broader pattern in human evolution: genes involved in immune function often show evidence of archaic introgression, presumably because variants that helped ancient populations survive local pathogens were advantageous when introduced into new populations.

More surprising was the discovery that ghost ancestry appears in genomic regions previously thought to reject all archaic DNA. These "archaic deserts" — stretches of the genome where Neanderthal and Denisovan ancestry is conspicuously absent — had been interpreted as regions where archaic variants were actively selected against, perhaps because they interfered with functions essential to modern humans.

TRACE found ghost ancestry in some of these deserts. This suggests a more nuanced story: these regions may have specifically rejected Neanderthal and Denisovan DNA, rather than all archaic ancestry. The ghost lineage, being more distantly related, may have carried variants that were less disruptive — or even beneficial — in these genomic contexts.

One such region contains FOXP2, a gene involved in speech and language development. The presence of ghost ancestry near FOXP2 does not mean the ghost lineage gave us language. It means that a stretch of DNA near a language-associated gene has an evolutionary history that does not match the Neanderthal or Denisovan contributions we already know about. Proximity is not function. But it is a flag for further investigation.

What the study does — and does not — say

The tempting headline is that scientists have "found two unknown human ancestors." It is understandable, memorable, and incomplete.

They have found statistical evidence that segments of contemporary DNA are best explained by admixture with deeply divergent populations that have not been directly sequenced. They have not:

Recovered a complete genome from an extinct individual.
Identified a fossil as the source of either signal.
Established that either lineage was a distinct species under any particular taxonomic definition.
Demonstrated that the ghost ancestry causes any specific modern human trait.

This may sound like a technical caveat. It is the whole story.

A useful way to frame it: TRACE is not a time machine. It is a forensic tool. It can say, this stretch of DNA has a genealogy that does not fit the lineages we have sampled. It cannot, on its own, say, this belonged to Homo heidelbergensis — or to any other fossil population.

The authors speculate that the more recent ghost signal could potentially correspond to a fossil-known population such as Homo heidelbergensis, a species that lived in Africa and Eurasia between roughly 700,000 and 200,000 years ago and is considered a plausible ancestor of both Neanderthals and modern humans. But this is a hypothesis, not an identification. The DNA signal and the fossil record have not been connected.

The research therefore changes the map before it fills in the labels.

Why this matters beyond a clever algorithm

Human origins research has been shaped by a preservation bias. Ancient-DNA discoveries have rightly transformed the field, but they also made fossil-rich, cold-region populations disproportionately visible. Neanderthals and Denisovans are famous partly because their DNA survived. Populations that lived in warmer climates — including much of Africa — left fewer molecular traces.

A method that can infer missing ancestry from contemporary DNA opens a route into demographic events that left no recoverable biological specimen. It allows researchers to test models of ancient African population structure that have remained difficult to examine with the fossil-DNA playbook alone.

That is the non-obvious connection: this is not merely an archaeology story. It is an access problem in science. For fields where direct evidence is unevenly distributed — because samples decay, institutions did not preserve them, or environments destroy them — better statistical reconstruction can alter which histories become legible.

It also shifts the scientific question. The previous question was often, "Which fossil gave us this ancestry?" The newer and more defensible question is, "What population history is required to produce this pattern, and which independent evidence could distinguish among competing histories?"

That is slower than declaring a new ancestor. It is also how the field avoids inventing certainty where there is only inference.

The hype check

There are four claims readers should decline to make from this study:

"Everyone has DNA from a newly discovered species." Not established. The inferred signal is widespread, but its biological source has not been directly observed or taxonomically classified.
"Scientists reconstructed the ghost ancestor's genome." Not yet. Because different people carry different fragments, the collective signal could eventually make more extensive reconstruction possible — but the study did not present a recovered ancient genome.
"This explains a modern human trait." Also not established. The signal appears near FOXP2 and in immune-related regions, but proximity is not a functional explanation. The study does not show that ghost ancestry caused a present-day cognitive, behavioural, or health trait.
"The ghost lineage was definitely Homo heidelbergensis." The authors mention this as a possibility. It is a hypothesis, not a finding.

The work is exciting precisely because it makes fewer claims than the viral version. It offers a method to ask better questions of the evidence we actually have.

Who is affected by the finding

Human-evolution researchers gain a tool that may reduce their dependence on the small set of places and fossils where ancient DNA survives. The next test is replication: different datasets, different demographic models, and comparisons with any future African ancient-DNA discoveries. If an ancient genome from Africa is eventually recovered, TRACE's predictions can be tested directly.

Museums and fossil researchers gain a more targeted way to link fossils, geography, and genomic predictions. TRACE cannot replace excavation; it can help decide which discoveries would be most informative.

The public gets a correction to a persistent but misleading picture of human ancestry. There was no clean hand-off from one neat species to the next. The history of Homo sapiens includes separation, contact, and admixture among populations whose remains are unevenly preserved. The better mental model is a braided stream, not a branching tree.

Commercial ancestry-testing customers should expect no immediate, consumer-level result. This is population-genetic research, not a new percentage on an ancestry dashboard. Any company implying otherwise would be moving well ahead of the science.

What this means for you

There is nothing practical to buy, test, or change because of this paper. The useful update is conceptual.

Treat neat diagrams of a single, branching human family tree with caution. The better mental model is a braided history: populations separated, met again, and exchanged ancestry repeatedly. Some of those populations are known from bones and DNA — Neanderthals, Denisovans. Others are currently visible only through the genetic traces they left in us.

For educators and science communicators, use the precise language: inferred archaic lineages, not "newly discovered species." That distinction preserves the discovery's real importance instead of shrinking it into a click-friendly fossil hunt.

Uncertainty ledger

Method validation: TRACE recovers known Neanderthal and Denisovan signals, but independent methods and datasets should test the unknown-lineage inferences. Replication is the next step.
Identity: Neither inferred lineage can currently be tied to a specific fossil population or geographic location with confidence. The Homo heidelbergensis hypothesis is plausible but unconfirmed.
Timing and admixture: Dates are model-based estimates, not calendar entries from a direct ancient sample. Different demographic models could produce different divergence and admixture dates.
Function: The study does not establish health, cognitive, or behavioural consequences of the identified segments. The FOXP2 proximity is intriguing but not explanatory.
The decisive evidence: Ancient DNA from currently under-sampled regions, especially Africa, could refine, corroborate, or overturn parts of the reconstruction — though preservation conditions make this difficult.
The desert-region finding: The presence of ghost ancestry in regions that reject Neanderthal and Denisovan DNA is a new observation. Its biological significance — whether these variants were neutral, beneficial, or simply less deleterious than Neanderthal/Denisovan variants — is unknown.

Bottom Line

This study did not discover a lost human species. It demonstrated that living genomes can preserve evidence of ancestral populations that the fossil record and ancient DNA have not yet captured — and it gave researchers a tool to read that evidence systematically. That makes TRACE important: it widens the evidence base for human origins without pretending it has solved the missing pieces. The map is larger now. The labels are still being written.

Sources

Tier 1: Zhang, Y., Biddanda, A., Johnson, S. A., O'Dushlaine, C., Moorjani, P. "Recovering signatures of archaic hominin introgression using ancestral recombination graphs." Science (30 July 2026). DOI: 10.1126/science.aef8874
Tier 1: Berkeley News — "New technique pinpoints human DNA inherited from 'ghost' ancestors" (30 July 2026)
Tier 1: ScienceDaily — "Scientists found two mysterious 'ghost' ancestors hiding in our DNA" (31 July 2026)
Tier 2: Live Science — "Scientists discover 2 new 'ghost' lineages that contributed DNA to modern humans" (30 July 2026)
Tier 2: The Debrief — "Scientists Discover Evidence of a Mysterious 'Ghost' Human Ancestor Hidden in Our DNA" (31 July 2026)
Tier 2: Sci. News — "Scientists Find Traces of Two Unknown Archaic Hominins in Human DNA" (2 August 2026)
Tier 2: Ars Technica — "Not just Neanderthals: Ghost lineage in Africa left its mark on our DNA" (31 July 2026)
Tier 2: Popular Archaeology — "TRACEing archaic hominin ancestry in modern humans without archaic genomes" (July 2026)