Epigenetic clocks & biological-age diagnostics

01Overview and value chain#

Markers EC: In Vitro Diagnostic Regulation (IVDR) Class C | OECD: Genomics & bioinformatics | Regulator: FDA (US), EMA (EU), NMPA (China)

Epigenetic clocks are machine-learning models that compute a person’s biological age from the methylation state of specific cytosine-guanine (CpG) dinucleotides scattered across the genome, exploiting the fact that DNA methylation drifts in a predictable, measurable pattern as cells age. A single Illumina Infinium MethylationEPIC BeadChip reads methylation intensity at more than 850,000 CpG sites simultaneously, and commercial reports now stack six or more clock algorithms — DunedinPACE, GrimAge 2.0, PhenoAge, OMICmAge and SymphonyAge among them — on top of the same array run. Consumer-facing tests range $229-799 per kit, turned around from a saliva swab or blood draw in two to four weeks. The sector spans four algorithm generations: first-generation clocks (Horvath, Hannum) trained to predict chronological age; second-generation clocks (PhenoAge, GrimAge) trained on clinical mortality and morbidity biomarkers instead; DunedinPACE, which measures the current pace of aging (e.g., 0.8 or 1.2 biological years per calendar year) rather than accumulated age; and emerging deep-learning clocks that fuse methylation with transcriptomic and proteomic layers.

The key directions of epigenetic-clock diagnostics are:

1. Multi-clock consumer panels: stacking several validated algorithms (DunedinPACE, GrimAge 2.0, PhenoAge, organ-specific ages) on one methylation-array run to give a single report several complementary age read-outs.
2. Pace-of-aging measurement: DunedinPACE-class algorithms trained on decades-long cohort studies to quantify the current rate of biological aging rather than an accumulated age number.
3. Alternative non-methylation biomarkers: glycan-based clocks that measure immunoglobulin G (IgG) glycosylation as a proxy for systemic inflammation and aging, avoiding the array/bisulfite pipeline entirely.
4. Geroprotector and intervention trials: using repeat epigenetic-age measurements as a primary or secondary endpoint to quantify whether a drug, therapeutic plasma exchange, or lifestyle change measurably slows or reverses biological aging.

Sectoral value chain#

[Sample collection (blood/saliva)] ──> [DNA extraction] ──> [Bisulfite/enzymatic conversion]
                  │                                         │                                      │
       (Stabilizing buffer)                    (Enzymatic protein digestion)                (EM-seq or bisulfite chemistry)
                                                                                                   │
  [Clinical report (bio-age)] <── [AI clock modeling (GrimAge/DunedinPACE)] <─── [Methylation array scan]

Value chain levels#

Cross-cutting technologies of the sector:

• Illumina Infinium MethylationEPIC BeadChip: the field’s reference array, reading methylation at over 850,000 CpG sites per sample via two-color fluorescent bead-based detection.
• Enzymatic methylation sequencing (EM-seq): an alternative to bisulfite conversion, which degrades up to 90% of input DNA; EM-seq uses sequential TET2 and APOBEC3A enzyme treatment to detect methylated cytosines while preserving DNA fragment length and structure.
• Elastic-net regression clocks: the DunedinPACE algorithm class, trained on the decades-long Dunedin Study cohort of about 1,000 participants followed from birth to age 45 across dozens of biomarkers, to quantify an individual’s current pace of aging.

02US#

The United States leads global commercialization of biological-age testing and the integration of epigenetic clocks into clinical trials of geroprotective compounds.

Steve Horvath’s original clocks, TruDiagnostic’s DunedinPACE license, Tally Health’s cheek-swab platform, Elysium’s PhenoAge-based Index#

• Origin of epigenetic clocks: UCLA human genetics professor Steve Horvath published the first multi-tissue epigenetic clock in 2013, showing biological age could be computed from methylation at 353 CpG sites; he now leads clock R&D at Altos Labs, alongside Yale-trained researcher Morgan Levine, who developed the second-generation PhenoAge algorithm before also joining Altos Labs.
• TruDiagnostic’s multi-clock platform: founded in 2020 and based in Lexington, KY, TruDiagnostic runs its TruAge COMPLETE and TruAge PACE reports off a single 900,000+ CpG methylation array, stacking DunedinPACE, GrimAge 2.0, PhenoAge, OMICmAge, SymphonyAge and 11 organ-specific age estimates in one $399-799 panel, positioning it as the largest commercial epigenetic-testing lab in the country.
• Tally Health’s cheek-swab CheekAge: founded in 2023 in New York and co-founded by longevity researcher David Sinclair, Tally Health runs a $229 cheek-swab test built on its proprietary CheekAge clock, publishing its methodology as a peer-reviewed scientific paper and pairing the age read-out with an AI-driven nutrition and supplement app.
• Elysium Health’s Index: founded in 2014 in New York, Elysium Health sells the $299 Index DNA-methylation test built on Morgan Levine and Steve Horvath’s 2018 DNAm PhenoAge algorithm, and separately markets a Senolytic Complex supplement, positioning the company across both the diagnostic and geroprotective-supplement sides of the longevity market.
• Buck Institute’s clock research: the Marin County, CA-based Buck Institute for Research on Aging published a new blood-based epigenetic clock in June 2025 focused on “intrinsic capacity” rather than raw age, and is running its BIG Study biobank to collect blood at scale for aging research; a May 2025 Aging Cell trial from the Institute found that therapeutic plasma exchange combined with IVIG reduced multi-omics biological age by 2.6 years on average.

03CN#

China is building population-scale epigenetic reference data and integrating biological-age testing into premium health-checkup packages, while its diagnostics sector adapts methylation-based cancer and aging assays to a domestic population whose methylation patterns differ from Western-trained clocks.

Population-specific clock calibration, oncology-diagnostics methylation platforms, premium health-checkup integration#

• Population-specific clock calibration: Chinese Academy of Sciences (CAS) institutes are building population-scale methylation reference datasets, since CpG methylation patterns carry ethnic and dietary variation that Western-trained clocks (built mostly on European and American cohorts) do not capture accurately.
• Methylation-based diagnostics platforms: domestic NGS diagnostics companies are extending DNA-methylation assays developed for early cancer detection toward biological-age and organ-aging scoring, reusing the same bisulfite/array infrastructure built for oncology screening.
• Premium health-checkup integration: private clinics in Beijing, Shanghai and Shenzhen are folding methylation-based biological-age scoring into annual premium health-checkup packages, the main commercial channel for the test in China today.

04EU#

The European Union anchors its biological-age diagnostics sector in long-running population cohorts, an alternative glycan-based biomarker pioneered in the UK and Croatia, and strict IVDR/GDPR oversight of genetic test data.

Long-running aging cohorts, GlycanAge’s IgG-glycosylation clock, IVDR Class C classification#

• Long-running population cohorts: European institutes coordinate decades-long longitudinal aging studies (such as Edinburgh’s Lothian Birth Cohorts), providing the deep genomic and epigenetic follow-up data used to train and validate new clock algorithms.
• GlycanAge’s glycan clock: London-headquartered GlycanAge, spun out of Croatian glycoscience research, markets a biological-age test based on immunoglobulin G (IgG) glycosylation rather than DNA methylation; a May 2026 study run jointly with Circulate Health analyzed roughly 20,000 individuals across 42 studies, establishing IgG glycans as an independent predictor of all-cause mortality and showing therapeutic plasma exchange can measurably reduce the glycan-based age score.
• IVDR Class C classification and GDPR: clinical deployment of biological-age tests in the EU falls under the In Vitro Diagnostic Regulation as a Class C genetic screening test, with storage and bioinformatic processing of the underlying genomic data separately bound by GDPR.

05Leading companies and research institutes#

06Tech stack and innovations#

The biological-age diagnostics stack pairs high-throughput molecular biology automation with cloud AI modeling:

1. Automated library prep on liquid-handling robots:
   • High-throughput DNA extraction and bisulfite/enzymatic conversion of up to 96 samples in parallel, with robotic liquid handlers dispensing reagents at 0.1 µL precision to eliminate cross-contamination between DNA samples.
2. Laser array scanning:
   • Illumina iScan-class scanners read EPIC BeadChips with dual high-intensity lasers, resolving methylated (one fluorescent channel) versus unmethylated (the other channel) cytosine signals across 850,000+ CpG sites per chip in minutes.
3. Cloud AI age-prediction models:
   • Regularized regression (ridge/lasso/elastic net) and deep neural networks trained on millions of CpG profiles compute age acceleration and pace-of-aging indices from the normalized beta-value array, generating a graphical predictive-risk report.

07Value chains and production pipelines#

Industrial pipeline for a clinical epigenetic-age test (ISO 15189)#

┌───────────────────────────┐      ┌───────────────────────────┐
│ 1. Saliva/blood collection │ ───> │ 2. Automated genomic DNA   │
│    in preservative tube    │      │    extraction              │
└───────────────────────────┘      └───────────────────────────┘
                                                 │
                                                 ▼
┌───────────────────────────┐      ┌───────────────────────────┐
│ 4. Array hybridization     │ <─── │ 3. Bisulfite/enzymatic     │
│    (Illumina EPIC BeadChip)│      │    conversion               │
└───────────────────────────┘      └───────────────────────────┘
              │
              ▼
┌───────────────────────────┐      ┌───────────────────────────┐
│ 5. Laser scan & beta-value │ ───> │ 6. AI clock scoring &      │
│    calculation              │      │    report generation       │
└───────────────────────────┘      └───────────────────────────┘

Stage 1: Sample collection and stabilization

A patient collects roughly 2 mL of saliva into a kit tube (e.g., a DNA Genotek Oragene-DNA collector); closing the cap punctures an internal membrane that releases a preservative buffer, lysing bacteria and inhibiting DNase enzymes so that genomic DNA in buccal cells stays stable at room temperature for up to five years during mail-in transport.

Stage 2: Automated genomic DNA extraction

The sample is logged by barcode and a 1 mL aliquot is dispensed by a robotic station into a 96-well plate. Proteinase K digestion at 50°C for one hour breaks down salivary proteins, and DNA is purified on silica-coated magnetic microbeads held by a magnet through alcohol wash steps, eluting at least 2 µg of DNA at an A260/A280 purity ratio of roughly 1.8.

Stage 3: Bisulfite or enzymatic methylation conversion

500 ng of genomic DNA undergoes bisulfite conversion (e.g., with a Zymo Research EZ DNA Methylation kit) or enzymatic methylation sequencing: a stepwise thermocycler profile denatures DNA at 95°C and sulfonates it at 50°C for up to three hours, converting unmethylated cytosines to uracil while leaving methylated cytosines (5-mC) intact; the converted single-stranded DNA is then purified and neutralized.

Stage 4: Array hybridization

The converted DNA undergoes whole-genome amplification for roughly 20 hours at 37°C, is enzymatically fragmented, precipitated, and resuspended in hybridization buffer, then loaded onto an Illumina Infinium MethylationEPIC BeadChip v2.0 and incubated in a hybridization chamber for about 20 hours at 48°C, allowing DNA fragments to hybridize to probe-bearing beads.

Stage 5: Single-base extension and laser scanning

The chip undergoes single-base extension with fluorescently labeled dideoxynucleotides, is stained, dried, and loaded into an Illumina iScan-class laser scanner, which reads fluorescence intensity across two channels (methylated versus unmethylated signal) and outputs high-density IDAT data files.

Stage 6: AI scoring and report generation

IDAT files are processed through a bioinformatics pipeline (commonly the R minfi package) for background normalization and noise filtering, computing a beta-value for each of 850,000+ CpG sites. The resulting beta-value vector is passed into GrimAge/DunedinPACE-class models, which compute biological age, the DunedinPACE pace-of-aging index, and disease-risk scores, generating an illustrated PDF report with longevity and geroprotector recommendations delivered to the patient’s portal.