Genomic context of the Horvath epigenetic clock

Research question

Do apparent clock-CpG enrichments persist when the comparison background better reflects platform availability?

Assess whether the 353 CpGs in the publicly documented Horvath 2013 clock have a distinctive genomic annotation profile compared with sites that could have been measured by the assay. In particular, compare the full Illumina 450K CpG background with a 27K-shared proxy closer to the clock's cross-platform selection opportunity. This is a compact annotation-based evaluation of Open-Science's research and reproducibility workflow, not a drug-discovery study or a validation of biological aging mechanisms.

Data and comparison design

The comparison uses the same clock sites against the full 450K set and the 27K-shared proxy. An enrichment odds ratio depends on how common an annotation is in the chosen background; platform selection can therefore mimic a biological preference.

Input or source	Role in this case	Source context
Horvath 2013 article	Clock definition and original candidate-universe context	Cite the article; the publisher's supplementary site list was not directly cross-checked in this run

Recorded findings

Annotation comparison	Recorded result	Why the background matters
Clock sites	353 CpGs	The fixed selected set.
Backgrounds	Full 450K and 27K-shared proxy	Platform availability changes the eligible comparison pool.
CpG-shore enrichment	Odds ratio 1.54 under shared proxy	Persists under that narrower comparison.
Island / promoter enrichment	Island no longer clear; promoter much smaller	Some apparent enrichment is background-sensitive.

Original analysis date: 2026-09-24; figure addition and new export: 2026-10-12. The original report compares 353 clock CpGs with 482,068 non-clock controls in the full 450K CpG assay and 25,625 non-clock controls in the 27K-shared proxy. It reports 72 feature/background comparisons, including chemistry-specific sensitivity analyses. The exact original 21,369-probe training-eligible universe was not recovered.

Against the full assay, CpG-island, shore and promoter odds ratios are 1.37, 2.36 and 3.77. Against the shared-platform proxy, they are 0.83 (no clear island association), 1.54 and 1.33. The report identifies shore enrichment as the most consistent residual feature: 147/353 clock sites (41.6%) versus 31.7% of shared-platform non-clock controls, OR 1.54 (95% CI 1.23–1.91). It explicitly limits these results to annotation/selection contrasts, not causal aging mechanisms.

What the comparison shows

Some annotation contrasts persist while others weaken with the background change. This makes the choice of eligible comparison sites central to the interpretation.

Assumptions and limits

The shared-platform set is a proxy, not the original training-eligible universe. Older annotations, correlated CpGs and unadjusted figure intervals limit interpretation. These annotation contrasts neither establish an aging mechanism nor clinically validate the clock.

Sources and research materials

The unchanged research package comes from upstream Case 2. It includes the archived session, analysis materials, results and figures. The poster uses an AI-generated illustrative photograph. This introduction summarizes archived findings; the analysis was not rerun.