7. The clock catalogue
pyaging’s registry is the most complete machine-readable inventory in existence: at commit e7267e9, 177 clocks with curated provenance (pyaging clock_metadata.json), backed by an evidence ledger of 6 author-confirmed, 332 code-confirmed, 2,435 paper-confirmed and 59 supplement-confirmed field assignments. This chapter describes that registry; FALCONAge’s own catalogue of 178 clocks is in the clock catalogue.
7.1 Distributions
Counted from pyaging’s clock_metadata.json at commit e7267e9 (177 clocks).
Data type. DNA methylation 156, histone modification 8, transcriptomics 6, clinical biomarkers 5, chromatin accessibility 2.
Species. Homo sapiens 157, multiple species 11, Mus musculus 7, C. elegans 1, human and mouse 1.
Platform. Illumina 450K 92, EPIC 87, 27K 13, MammalMethylChip40 10, ChIP-seq 8, RNA-seq 6, clinical laboratory assays 5, RRBS 3, expression microarray 3, mammalian methylation array 2, MammalMethylChip320 2, ATAC-seq 2, bisulfite sequencing 2, targeted bisulfite sequencing 1.
Model type. Elastic net 66, LASSO 18, reference-based constrained deconvolution 18, PCA + elastic net 18, PCA + elastic net + ARD 8, linear regression 5, weighted linear score 3, causality-weighted elastic net 3, and two each of logistic regression, single-CpG score, PCA + Cox, Klemera–Doubal composite, mean methylation aggregation, dynamic methylation transmission, elastic net Cox, 95th-percentile aggregation and PCA. One each of: deep neural network, Cox proportional hazards, elastic net Cox ensemble, bootstrap-stabilised elastic net, complement of mean methylation, two-stage elastic net + Cox, feature-selected logistic, Mahalanobis distance composite, mean aggregation, two-stage elastic net, ridge logistic, orthologue-transferred ridge logistic, penalised hazards with Gompertz calibration, Gompertz hazards with age calibration, weighted methylation aggregation, ridge, elastic net logistic, quadratically calibrated elastic net, screened elastic net and BLUP.
Unit. Years 86, proportion 20, unitless 18, pg/mL 7, beta value 6, weeks 6, months 4, probability 3, and two each of biological years per chronological year, days, cell divisions per stem cell, pack-years, ratio, log odds, population doublings and log hazard. One each of hours, passages, m/s, kg, mL/kg/min, kb, relative age, ln(%), ln(mg/L), units per week and bp.
Population. Adults 78, older adults 29, all ages 27, multiple mammalian species 11, human age unspecified 7, mice 6, newborns 4, human cell cultures 4, pregnancies 4, and one each of C. elegans, centenarians, humans and mice, children and adolescents, adult men, prenatal and newborn, and children.
7.2 Selected clocks with their key parameters
The full 177-row table is in the registry; this is the subset a new tool should support first, sorted by likely usage.
| Clock | Year | Layer | Features | Model | Predicts | Unit | Notes |
|---|---|---|---|---|---|---|---|
| horvath2013 | 2013 | DNAm multi-tissue | 353 | elastic net | CA | years | anti_log_linear; the reference clock |
| hannum | 2013 | DNAm blood | 71 | elastic net | CA | years | |
| lin | 2016 | DNAm blood | 99 | linear | CA | years | intercept 12.2169841 |
| skinandblood | 2018 | DNAm skin/blood/fibroblast | 391 | elastic net | CA | years | most accurate first-gen across tissues |
| phenoage | 2018 | clinical chem | 10 | Gompertz-calibrated penalised hazards | phenotypic age | years | see §5.1 |
| dnamphenoage | 2018 | DNAm blood | 513 | elastic net | phenotypic age | years | |
| zhangen | 2019 | DNAm blood/saliva | 514 | elastic net | CA | years | per-sample z-score |
| zhangblup | 2019 | DNAm | 319,607 | BLUP | CA | years | per-sample z-score |
| grimage | 2019 | DNAm blood | 1,032 | 2-stage EN + Cox | mortality | years | 8 sub-clocks |
| grimage2 | 2022 | DNAm blood | 1,032 | EN Cox | mortality | years | 10 sub-clocks (+logCRP, +logA1C) |
| dunedinpoam38 | 2020 | DNAm blood | 46 | elastic net | pace | bio-y/chron-y | |
| dunedinpace | 2022 | DNAm blood | 20,000 (173 scoring) | elastic net | pace | bio-y/chron-y | quantile norm required |
| altumage | 2022 | DNAm multi-tissue | 20,318 | DNN | CA | years | 5 hidden layers, SELU |
| pchorvath2013 / pchannum / pcphenoage / pcgrimage / pcskinandblood / pcdnamtl | 2022 | DNAm | 78,464 (78,466 for PCGrimAge) | PCA + EN | various | years / bp | high test-retest ICC |
| hrsinchphenoage | 2022 | DNAm blood | 959 | weighted linear score | phenotypic age | years | |
| yingcausage / yingdamage / yingadaptage | 2024 | DNAm blood | 585 / 1,089 / 999 | causality-weighted EN | CA / damage / adaptation | years | MR-filtered CpGs |
| dnamfitage | 2023 | DNAm blood | 630 | Klemera–Doubal composite | fitness bio-age | years | sex-specific |
| dnamtl | 2019 | DNAm blood | 140 | elastic net | telomere length | kb | |
| stemtoc | 2024 | DNAm multi-tissue | 371 | 0.95-quantile aggregation | mitotic age | beta | |
| epitoc1 / epitoc2 / epitoc3 | 2016/2020/2020 | DNAm | 385 / 163 / 170 | mean / transmission model | mitotic age | beta / divisions per stem cell | |
| systemsage + 11 system variants | 2025 | DNAm blood | 125,175 | PCA + EN | 11 system ages + composite | years | |
| cpgptgrimage3 | 2024 | DNAm-derived proxies | 24 | Cox | mortality/bio-age | years | needs CpGPT + GrimAge2 proxies |
| cpgptpcgrimage3 | 2024 | DNAm-derived proxies | 31 | PCA + Cox | mortality/bio-age | years | |
| intrinclock | 2024 | DNAm multi-tissue | 380 | 2-stage EN | CA | years | stable across immune cell types |
| encen40 / encen100 | 2023 | DNAm blood/saliva/buccal | 559 / 198 | elastic net | CA | years | centenarian-tuned |
| retroelementagev1/v2 | 2024 | DNAm blood | 1,317 / 1,378 | elastic net | CA | years | retroelement-focused |
| senchronoage / sencultureage / senmortalityage | 2026 | DNAm | 187 / 142 / 91 | EN / EN-logistic / EN-Cox | CA / senescence / mortality | years / log-odds / log-hazard | newest human clocks in the registry |
| pasta / reg | 2025 | transcriptomics multi-tissue | 8,113 | ridge-logistic / ridge | transcriptomic age | years | rank-normalised |
| bitage | 2021 | transcriptomics C. elegans | 576 | elastic net | bio age | hours | binarised |
| ocampoatac1/2 | 2023 | ATAC-seq PBMC | 228 / 380 | elastic net | CA | years | |
| camillo* (8 marks) | 2025 | histone ChIP-seq | 102–3,739 | PCA + EN + ARD | CA | years | |
| mammalian1/2/3 + blood/skin variants | 2023 | DNAm mammalian array | 335–2,572 | elastic net | CA | years | species lookup required |
| petkovich / stubbs / meer / thompson | 2017–18 | DNAm mouse | 90 / 17,992 / 435 / 582 | elastic net | CA | months / days | RRBS or bisulfite-seq |
| deconvolute* (6 + 12 cell) | 2018 / 2022 | DNAm EPIC | 600 / 240 | constrained deconvolution | cell proportions | proportion | |
| xchrom / ychrom | 2021 | DNAm | 4,047 / 284 | PCA | sex chromosome dosage | unitless | |
| mccartney* (10) | 2018 | DNAm blood | 204–1,109 | LASSO | lifestyle/clinical traits | various | smoking, BMI, alcohol, lipids, education, WHR, body fat |
| knight / bohlin / mayne / lee×3 / epicga | 2016–2021 | DNAm cord blood/placenta | 62–558 | EN / LASSO | gestational age | weeks | |
| pedbe / wu | 2019–2020 | DNAm buccal/blood | 94 / 111 | EN | CA in children | years | |
| corticalclock / neusin / gliasin | 2020 / 2024 | DNAm brain cortex | 347 / 672 / 220 | EN | CA | years | |
| ctsliver / hep / hepatoxu | 2017–2024 | DNAm liver / cfDNA | 90 / 70 / 10 | LASSO / logistic | CA / HCC | years / unitless | |
| cvdwesterman / adbahadosingh / prostatecancerkirby / depressionbarbu / downsyndrome | 2017–2021 | DNAm blood | 235 / 4 / 3 / 196 / 652 | disease risk models | disease probability or score | probability / log odds / unitless | |
| dnamfili / dnamic / dnamstress | 2022–2025 | DNAm blood | 20 / 91 / 211 | LASSO / EN | frailty / intrinsic capacity / stress | unitless | |
| replitali / replitalinorm / cellpopage | 2022–2024 | DNAm cultured cells | 87 / 218 / 42 | EN | replicative history | population doublings / passages |
7.3 Clocks outside pyaging worth supporting
- organAging - 24 organ panels × {chronological, mortality} × {full Olink 3072, feature-reduced Olink 1536, CSF}. Coefficients ship as CSVs in the repo.
- tAge - 60 transcriptomic models (§4.18).
- SpatialAgingClock - 18 mouse brain cell types (§4.13).
- scImmuAging - 5 human PBMC cell types (CD4T, CD8T, MONO, NK, B), glmnet objects with
lambda.min, applied to pseudocells. - SCALE - >20 mouse tissues, Tabula Muris Senis.
- CellBiAge - mouse brain snRNA/scRNA binary classifiers.
- scAge / scEpiAge - single-cell methylation.
- OMICmAge - 421 epigenetic biomarker proxies (110 protein, 286 metabolite, 25 clinical) distilled into a DNAm-only predictor of EMRAge. Discovery MGB Aging Biobank n = 3,451; validation TruDiagnostic n = 14,213 and Generation Scotland n = 18,672.
- GOLD BioAge / GOLD ProtAge / GOLD MetAge / Light BioAge - Gompertz-law clinical, proteomic, metabolomic, and 3-marker variants (Adv Sci 12(32), 2025); benchmarked on NHANES and UK Biobank.
- StackAge - UKB proteomics + metabolomics ensemble, n = 30,376.
- Oh 2023 organ clocks - SomaScan, ~5,700 people,
organagePython package. - GP-age - Gaussian process regression on 30 CpGs selected by k-means over 80 age-correlated sites; MedAE 2.1 y, 11,910 blood samples.
- RNAAgeCalc - cross-tissue and tissue-specific GTEx transcriptomic ages (R/Bioconductor, Python
racpy, Shiny). - MEAT - muscle epigenetic age test (Bioconductor).
- MiAge - mitotic age; coefficients ship in ComputAge.
- GlycanAge, IMM-AGE, microbiome clocks.
Of the clocks outside pyaging, FALCONAge now scores MetaboAge and MetaboHealth (NMR metabolomics), LinAge2 and cAge; OMICmAge is catalogued as licensed.