Computational Longevity
Mapping the architecture of cellular aging.
Cartora Labs builds the computational foundation for senescence — mapping the hidden architecture that lets aging cells evade the body's defenses, and working toward a blood readout of the burden they leave behind.
01 — The discovery
Aging cells learn to hide. We found the signature.
As we age, cells stop dividing but refuse to die — senescent cells. They linger in our tissues and leak a toxic broth of inflammatory signals. The immune system is supposed to clear them. Many aren't cleared — and we found a specific molecular signature that tracks the cells doing the evading.
Across millions of single immune cells, the CD8 T cells that accumulate senescence markers with age share a common trick: an immune-evasion keystone — a paired "don't-eat-me / don't-kill-me" program built on two molecules, CD47 and LGALS1. It is the architecture that lets a broken cell talk its way past the macrophages and natural-killer cells sent to remove it.
The pair is specific, not incidental. Tested against thousands of expression-matched random gene pairs, CD47 + LGALS1 carries age information those pairs do not — it passes that null in two independent cohorts. On its own it accounts for the majority of the age signal in CD8 T cells, out-tracking the classic senescence core. That specificity is the core of our patent-pending finding.
From it we are building a small, blood-readable panel scored on CD8 T cells from a routine draw — aimed at a measure of senescent burden rather than a birthday. The panel is in dry-lab development; wet-lab confirmation is the next gate, not a finished result.
Burden, not an age clock
The goal is not a number that only goes up. We are building toward a readout of how much senescent burden a body carries now — one that could be acted on and re-measured. That claim is the target of our next gate, not a result we hold today.
The immune-evasion keystone
CD47 + LGALS1 is a shared strategy senescent cells use to escape clearance. Remove it in silico and the age signal drops sharply in two of three native cohorts — evidence the pair is load-bearing, though causality is not established.
Readable from a blood draw
A compact gene/protein panel scores each cell, then counts the high-scoring fraction per donor. Designed to run on standard flow cytometry — no exotic reagents.
02 — How it works
A per-cell score, then a count.
Most aging biomarkers average a whole sample into one bulk number and lose the cells that matter. We do the opposite: score every cell for the senescence signature, set a threshold against a young reference, and count the fraction that crosses it. Counting cells, not averaging them, is the architecture we build on.
The panel combines a senescence core (a cell-cycle-arrest marker up, a nuclear-envelope marker down) with the immune-evasion keystone (CD47 + LGALS1 up). Each cell gets a score; the high-scoring fraction per donor is what tracks age — robustly, across ethnicities, both sexes, and independent of CMV infection status.
A subtle but important detail we uncovered: the classic arrest gene p21 behaves backwards in blood T cells compared to fibroblasts — so we drop it from the blood panel and keep it only for tissue. Compartment-specific biology, replicated across cohorts and the subject of our preprint.
What holds — and what's open
Accumulated immune age
The high-scoring fraction of cytotoxic CD8 T cells rises steadily across the decades of life, in every cohort we have tested. CD8 is where the signal is specific and reproducible — so CD8 is the whole of our lead. Other compartments were tested and scoped out.
What we have not shown yet
Tracking age is not the same as counting senescent cells. Against directly β-gal-sorted CD8 cells, our two up-markers move the right way in 13/13 donor pairs — but the frozen composite does not yet clear the full ground-truth gate. Closing that gap is the current program, and we would rather say so than call it validated.
03 — The evidence
Eight cohorts. One signal.
The signature was developed on a gold-standard atlas and then held to independent cohorts spanning six ethnicities, both sexes, and four continents — plus 53,000 plasma proteomes. On the age-tracking endpoint, every cohort points the same way.
| Cohort | Origin | Donors | Correlation |
|---|---|---|---|
| OneK1K | Australia — gold-standard atlas | ~1,900 | +0.49 · permutation z ≈ 15σ |
| CIMA | Asia — largest replication | 421 | +0.562 · p = 2×10⁻³⁶ |
| KCL | UK — CMV-stratified | 188 | +0.436 · p = 4×10⁻¹⁰ |
| Terekhova | USA — independent, powered | 166 | +0.390 · p = 2×10⁻⁷ |
| AIDA | Asia — 6 ethnicities | 595 | +0.435 · cross-ethnicity transfer |
| UCSF | USA — 97% female | 99 | +0.348 · p = 4×10⁻⁴ |
| + Allen (USA) and Nehar-Belaid cohorts — all eight independent cohorts positive. | |||
The signal survives the three standard attacks on an aging biomarker: it is not merely compositional (the pooled signal is affected by the naive-to-memory shift, but it stays above its matched null when the canonical CD8 states are equally weighted), sex-transferable (strong in powered female cohorts), and CMV-robust (equal in CMV-positive and CMV-negative donors).
One honest caveat, stated plainly: every number above is age-tracking. Correlation with chronological age is necessary but not sufficient — it is not by itself proof that we are counting senescent cells. That is what the wet-lab gate is for.
04 — Applications
Where a map of senescent burden matters.
Senescent cells accumulate wherever we age — the immune system, the vasculature, the skin. A quantitative, re-measurable readout of senescent burden would have uses from the clinic to the cosmetic bench. Below is where we are aiming, and how far along each direction actually is.
Blood biomarker
Score the senescence-associated CD8 fraction from a routine draw — intended as a burden readout to baseline healthy aging and flag elevated load. This is the lead program.
Dry-lab · age-tracking shownMonitoring senescence therapy
A per-cell burden score is, by construction, re-measurable after an intervention — a way to ask whether a senolytic actually lowered the burden. Demonstrating that a score moves with treatment is future work, not a current result.
Wet-lab pilot · next gateCardiovascular risk
In our tissue analyses the aorta carries the heaviest load of inflammatory-signal receptors — consistent with the field's view that vascular aging is a downstream consequence of systemic senescent toxicity.
Research directionSkin & visible aging
Skin is where senescence is most visible: senescent dermal fibroblasts and their inflammatory secretions are a well-established driver of photoaging in the wider literature. A burden readout is a natural fit for measuring and monitoring dermal senescence — a direction the platform is built to extend toward.
Vision · field biology05 — Intellectual property
Filed, published, and reproducible.
The core invention — scoring senescence-associated CD8 T cells via the CD47 + LGALS1 immune-evasion keystone — is covered by a US provisional application and published as an open preprint with a permanent DOI.
-
Novelty anchor. LGALS1 as a blood CD8 senescence marker, paired with CD47, is our novelty zone — distinct from existing single-marker immune-age tools. CD47 alone is prior art and we claim no rights to it. No freedom-to-operate opinion is asserted here.
-
Specific, not incidental. The pair carries age information that thousands of expression-matched random gene pairs do not, in two independent cohorts — and it adds information orthogonal to the classic p16 marker.
-
Ablation-supported. Removing the keystone drops the signal sharply in 2 of 3 native cohorts; in the third the senescence core carries it alone. Evidence the pair is load-bearing — reported with its exception rather than rounded up.
-
Fully reproducible. Methods and results are published openly under a permanent DOI so any lab can re-derive the finding.
06 — Contact
Quietly building. Open to the right partners.
For research collaboration, wet-lab / CRO partnership, licensing, or investment — we'd like to hear from you.