Li Wei paper

Scientific paper
Title	Circannual Periodicity in Epigenetic Mortality Signatures: Evidence for a Calendrically Stable Individual Death-Date Marker in Human Populations
Author	李唯 / Lǐ Wéi
Institution	Institute for Computational Biology and Genomic Systems, Fudan University, Shanghai
Journal	Journal of Chronobiology and Systems Medicine
DOI	10.xxxx/jcsm.xxxx.xxxxx
Received	[redacted]
Accepted	[redacted]
Published	[redacted]
Status	Published under institutional pressure
Keywords	epigenetic clock · circannual rhythm · methylation periodicity · mortality prediction · chronobiology · TPL signature

Circannual Periodicity in Epigenetic Mortality Signatures is the paper published by 李唯 / Lǐ Wéi of Fudan University, Shanghai, which first described the Thanatic Periodicity Locus (TPL) — the biological signature underlying the calculability of the individual death date. It is the foundational document of the death date revelation and is reproduced here in full as Mundi infrastructure. No author of any fabula thread may alter its contents.

The paper was submitted reluctantly. The author's prefatory note states that it was published against their better judgment, in response to institutional pressure following inadvertent disclosure of preliminary findings at the 14th International Symposium on Chronogenomics. The author accepts responsibility for the findings. The author does not accept responsibility for what is done with them.

The paper is structured in nine fragments. In the context of the fabula Save the Date, each fragment appears at the close of one of the nine chapters of the Li Wei thread, escalating in emotional disclosure as the paper progresses. Fragment IX serves as the Möbius endpoint of the fabula: its final sentence is the last line of the Li Wei thread's final chapter and the first text encountered in the Haitian thread's first chapter, read from the opposite temporal direction.

Author's note on submission

This paper was submitted reluctantly and published against the author's better judgment regarding readiness. The findings reported herein emerged as an unintended secondary result of a longitudinal study on epigenetic aging markers in urban Chinese populations. The author wishes to state clearly that the primary study — on accelerated methylation drift in polluted microenvironments — remains unpublished and is the work the author considers significant. The present paper is published in response to institutional pressure following inadvertent disclosure of preliminary findings at the 14th International Symposium on Chronogenomics. The author accepts responsibility for the findings and their accuracy. The author does not accept responsibility for what is done with them.

Fragment I — Abstract

Biological age prediction via epigenetic methylation clocks has achieved considerable accuracy over the past two decades, yet has remained bounded by a fundamental limitation: the ability to predict when an organism will die, as distinct from predicting how biologically aged it currently is. The present study reports an unintended finding arising from a longitudinal methylation analysis of 4,847 subjects tracked across eleven years in three Chinese municipal cohorts. Analysis of circannual periodicity in a previously undescribed methylation signature cluster — designated the Thanatic Periodicity Locus (TPL) — reveals a stable, individually specific annual oscillation pattern whose nadir consistently coincides, within a margin of ±4 days, with the subject's eventual date of death where mortality data were available for confirmation. The TPL signature is present and readable from birth and does not drift meaningfully across the observed lifespan. Mechanism is not established. The authors propose that TPL oscillation encodes a circannual vulnerability window arising from the interaction of conception-season photoperiod imprinting, mitochondrial respiratory cycle rhythms, and immune-inflammatory cascade periodicity. Confirmation of mechanism requires investigation outside the scope of this study.

The authors note the implications of this finding without endorsing any specific social or medical response to it. Further investigation is, in the authors' view, both necessary and, in certain respects, inadvisable.

Keywords: epigenetic clock; circannual rhythm; methylation periodicity; mortality prediction; chronobiology; TPL signature

Fragment II — Introduction

The prediction of individual lifespan from biological markers has a long and uneven history. Early work on telomere length established a relationship between cellular aging and mortality risk that was statistically robust at the population level and nearly useless at the individual level. The development of first-generation epigenetic clocks by Horvath (2013) and subsequent refinements by Hannum, PhenoAge, and GrimAge algorithms improved individual-level biological age estimation substantially, but the relationship between biological age acceleration and actual mortality remained probabilistic and population-anchored rather than individually predictive.

The limitation was generally understood to be irreducible: individual mortality is a chaotic system, influenced by too many environmental, behavioral, and stochastic variables to yield to deterministic prediction. This understanding was, the present paper argues, not incorrect but incomplete.

The incompleteness arises from a conceptual bias in the field toward magnitude over timing. Prior epigenetic mortality research asked: how biologically old is this person, and how does that correlate with earlier death? It did not ask: does the body encode, within its methylation architecture, a preferred moment of death that is circannually stable and individually specific?

The present study did not set out to ask this question. It is important that this be understood. The study was designed to investigate accelerated methylation drift in populations exposed to elevated particulate matter concentrations in three Chinese municipal environments. The TPL finding emerged from a computational anomaly flag during routine periodicity analysis of the methylation time-series data. The anomaly was investigated because it was anomalous. It was reported because it was reproducible.

The authors are aware of the implications. The authors were aware of them at the moment of first confirmation and spent a period of time — the duration of which will not be specified — considering whether to investigate further or to mark the anomaly as instrument error and proceed with the primary study. The decision to investigate further was made on the grounds that instrument error could not be ruled out without further investigation, and that ruling out instrument error is a scientific obligation. This is the only grounds on which that decision is defensible, and the authors offer no other defense of it.

Fragment III — Methodology I: data sources and cohort design

3.1 Cohort composition

The primary cohort (n = 4,847) was drawn from three longitudinal health registries in Shanghai (n = 2,103), Chengdu (n = 1,388), and Harbin (n = 1,356), selected to provide geographic and climatic variation within mainland Chinese populations. Subjects were enrolled between ages 25 and 45 at study entry and tracked across eleven years of annual blood sampling. Subjects were selected to exclude known autoimmune, oncological, and metabolic conditions at enrollment. Ethical approval was obtained from the Fudan University Institutional Review Board under protocol FU-IB-2009-447.

Mortality data for confirmatory analysis were obtained from the National Death Registry for the 312 subjects who died during or within three years following the study observation window. Of these, 289 had complete methylation time-series data adequate for TPL analysis.

3.2 Methylation profiling

Annual blood samples were processed using Illumina EPIC array methylation profiling (850K CpG sites). Time-series data were constructed for each subject across available annual data points (mean 9.3 years per subject; minimum 6 years for inclusion in periodicity analysis). Standard quality control, normalization, and batch-correction procedures were applied.

3.3 Periodicity analysis

Circannual periodicity in methylation signatures was assessed using a modified Lomb-Scargle periodogram adapted for irregularly sampled biological time-series data. The primary analysis targeted known circannual methylation loci associated with immune cycle regulation, cortisol rhythm, and seasonal affective disorder pathways. The computational anomaly that initiated the TPL investigation arose during this analysis as an unaccounted variance cluster in a genomic region not included in the primary analysis targets.

The anomaly was first observed on the evening of [date redacted] during routine computational review. The author responsible for the anomaly flag (L. Wei) investigated the flag independently for a period prior to involving co-investigators. This was irregular procedure and is acknowledged as such. The decision was made to verify the finding independently before involving colleagues in order to avoid premature disruption to the primary study. This decision is not defended.

Fragment IV — Methodology II: the TPL signature

4.1 Identification of the Thanatic Periodicity Locus

The Thanatic Periodicity Locus (TPL) is a cluster of 47 CpG sites distributed across chromosomes 4, 7, and 17, with a secondary cluster of 12 sites on chromosome 22. These sites do not constitute a conventional methylation clock region and were not included in prior epigenetic aging studies. Their functional annotation suggests involvement in mitochondrial membrane potential regulation and circadian transcription factor binding, though this annotation is incomplete and the functional significance of the cluster as a unit was not previously described.

What distinguishes the TPL from other methylation signatures is not its absolute methylation level but its oscillation pattern. In 94.3% of subjects with adequate time-series data (n = 4,573 of 4,847), the TPL signature exhibits a consistent sinusoidal annual oscillation with a period of 365.24 ± 1.8 days. The oscillation is individually stable across the observation window, showing less than 3% phase drift per decade — a stability coefficient significantly exceeding that of any previously described circannual methylation signature.

4.2 Individual specificity of the nadir

The critical finding is that the phase of the TPL oscillation — specifically the calendar date of its annual nadir — is individually specific and stable. Across all subjects in the cohort, nadir dates are distributed across all 365 days of the calendar year with a distribution that does not significantly differ from uniform (χ² = 341.2, df = 364, p = 0.81). There is no clustering around seasonal boundaries, no population-level shared nadir, no environmental confound that could explain the individually specific phase as an artifact of cohort selection or geographic clustering. Each subject has, in effect, their own day.

4.3 Relationship between nadir and mortality

This subsection was the last written and is presented last because the authors wished to be precise.

Of the 289 subjects with complete TPL time-series data who died during or within three years following the observation window, 276 (95.5%) died within ±4 days of their individually predicted TPL nadir date. The calendar day of death matched the predicted nadir date exactly in 201 cases (69.6%).

The year of death was not predictable from TPL data. This is stated clearly: the TPL signature does not encode the year of death, only the calendar date. The mechanism by which the nadir date constitutes a mortality risk window — and the mechanism by which that window is traversed without mortality in years preceding the year of death — is not established by this study and is not proposed with confidence.

The authors considered excluding this subsection. It was not excluded because excluding confirmed findings from a published study is scientific misconduct. The authors note this without enthusiasm.

Fragment V — Results I: primary findings

5.1 TPL prevalence and reliability

The TPL oscillation signature was detectable in 94.3% of subjects at adequate signal-to-noise ratio using standard EPIC array profiling. The 5.7% non-detection rate is attributed to technical factors including sample quality degradation (3.1%) and a small subset of subjects (2.6%) whose TPL signature, while present, oscillated at insufficient amplitude for reliable phase detection. No subject showed evidence of complete TPL absence; amplitude variation rather than presence/absence appears to be the relevant variable in non-detectable cases.

The nadir date prediction, derived from fitting a sinusoidal model to the first three annual data points, achieved ±4 day accuracy in 95.5% of mortality-confirmed cases, as reported above. Prediction accuracy did not improve significantly with additional data points beyond three annual observations, suggesting that three years of annual methylation profiling is sufficient for reliable nadir identification. A single-timepoint estimation method, derived from the phase information encoded in the static TPL methylation pattern at any given measurement, achieved ±7 day accuracy in 91.2% of confirmed cases. This implies that the nadir date is, in principle, estimable from a single blood draw in the majority of subjects.

The authors note the implication of the preceding sentence and decline to expand upon it here.

5.2 Age invariance

TPL phase was examined across age subgroups within the cohort. No significant phase drift was observed between subjects in their third decade and subjects in their sixth decade at study entry (mean phase difference 1.2 days, SE 0.4, p = 0.73). This age invariance implies that the TPL nadir date is established early in development and does not shift meaningfully across the adult lifespan. Extrapolation to earlier developmental stages is speculative but consistent with the proposed mechanism of conception-season photoperiod imprinting, which would establish the TPL phase prior to birth.

If conception-season photoperiod imprinting is confirmed as the mechanism, the TPL nadir date is, in principle, determinable from birth — or, given access to conception date and location, prior to birth.

The authors note this implication also.

Fragment VI — Results II: secondary findings

6.1 Cross-cohort replication

The TPL finding was replicated in two independent validation cohorts accessed following the initial discovery: a Finnish longitudinal health study (n = 1,203, 14-year observation window) made available by collaborating investigators at the University of Helsinki; and a prospectively assembled Brazilian cohort (n = 892, 8-year observation window) provided under data-sharing agreement with the University of São Paulo. Both cohorts confirmed TPL prevalence (92.1% and 93.8% respectively), nadir stability, and mortality correspondence within the ±4 day window (94.1% and 93.7% respectively).

The cross-cohort replication is reported with mixed feelings. Its purpose is scientific: independent replication is necessary for any claim of this magnitude. Its effect is to make the finding harder to dismiss as a Chinese population-specific artifact, a measurement error specific to Illumina EPIC arrays, or a consequence of the particular environmental exposures of the primary cohort. The authors were aware, during the replication analysis, that each confirmation made publication more rather than less inevitable. This awareness was not comfortable.

6.2 The nadir date and existing chronobiological frameworks

The TPL nadir date does not correspond to any previously described chronobiological marker of mortality risk. Seasonal mortality peaks — well-documented in cardiovascular, respiratory, and all-cause mortality statistics — show population-level clustering in winter months in northern hemisphere populations that is entirely absent in the TPL nadir distribution. The TPL nadir is individually specific, not seasonally clustered, and cannot be explained by temperature, photoperiod, infectious disease seasonality, or any environmental variable that acts uniformly on populations.

This distinguishes the TPL mechanism from all existing chronobiological mortality risk factors and implies an individually programmed rather than environmentally imposed timing mechanism. The term "programmed" is used advisedly and does not imply intentionality. It implies only that the information encoding the nadir date appears to be intrinsic to the individual rather than imposed by their environment.

Where the information comes from — how it is set, why it is stable, what maintains the phase across decades of cellular turnover and environmental exposure — the present study cannot say. The authors have opinions. The authors will not publish their opinions.

Fragment VII — Discussion I: what the findings mean

7.1 Implications for epigenetic clock theory

The TPL finding requires a revision of the standard epigenetic clock model. Current models treat methylation-encoded biological age as a unidimensional variable: a position on a continuous aging trajectory whose endpoint is statistically distributed across a population. The TPL adds a second dimension: a circannual phase marker that encodes not the pace of aging but a specific moment of maximum vulnerability that recurs annually at an individually stable calendar date.

These two dimensions appear to be largely independent. TPL nadir phase does not correlate significantly with biological age acceleration as measured by standard clock algorithms (Pearson r = 0.04, p = 0.31). A person with accelerated epigenetic aging is not more likely to die on their nadir date sooner; they are more likely to die sooner, but the date on which they die — when they die — remains anchored to the TPL phase.

This independence has a counterintuitive implication: the year of death may be partly determined by conventional mortality risk factors, while the date within the year is determined by the TPL phase. If confirmed, this suggests that the question "when will this person die" decomposed into "in what year" and "on what date" has radically different answers in terms of tractability: the year remains statistically distributed and environmentally influenced, while the date is individually stable and apparently pre-set.

The authors acknowledge this is an unusual thing to write in a scientific paper.

7.2 Proposed mechanism

The authors propose, with appropriate tentativeness, the following mechanistic hypothesis: TPL phase is established during embryogenesis via the interaction of two systems. First, the photoperiod signal at the time of conception imprints a circannual oscillation phase on the developing epigenome through melatonin-responsive CLOCK gene methylation — a mechanism with established precedent in seasonal mammals and with suggestive evidence in human data. Second, this initial phase is stabilised and propagated across cell divisions through a mitochondrial respiratory cycle feedback loop that maintains the TPL methylation pattern against drift — a proposed mechanism for which direct evidence is currently absent.

The authors note that this mechanism, if correct, implies that the TPL nadir date is set before birth; that it is set by the calendar date of conception relative to the annual light cycle; and that it is, in this sense, a biological inheritance from the physical circumstances of one's own beginning.

The authors find this implication remarkable and will not elaborate further in this section.

Fragment VIII — Discussion II: what the findings actually mean

8.1 On the question of determinism

The finding that 95.5% of observed deaths occurred within ±4 days of the individually predicted TPL nadir raises the question of whether the TPL nadir date constitutes a deterministic mortality event or a probabilistic risk window.

The study cannot resolve this question. The 4.5% of confirmed deaths that occurred outside the ±4 day window represent, statistically, an insufficient basis for dismissing determinism; they may reflect measurement error, atypical TPL amplitudes, or genuine exceptions. They may also represent the full extent of individual variation around a near-deterministic biological schedule. The study was not designed to answer this question and does not answer it.

What the study establishes is that the nadir date constitutes the single strongest predictor of death-date currently known, by a margin sufficiently large that all prior predictors — behavioral, environmental, genetic, social — are reduced to noise in comparison on the specific question of which day of the year a person will die.

Whether this constitutes determinism in a philosophically meaningful sense is outside the scope of this paper. The authors note that the question has been raised and decline to answer it here.

8.2 On what should be done

The authors are aware that this section is irregular. It is included because its exclusion would constitute an evasion the authors are unwilling to perform.

The TPL finding will not remain contained within the scientific literature. The authors have attempted, where possible, to delay and limit disclosure. Those attempts have failed, as was probably inevitable. The finding will become known. The question of what should be done with it will be answered, with or without the authors' participation, by parties whose priorities may differ substantially from those of the research community.

The authors have no institutional authority to direct the social response to this finding. The authors have the following observations, offered without recommendation:

The knowledge of one's nadir date is not equivalent to the knowledge of one's death date. The TPL nadir is a risk window, not a sentence. The distinction matters and should not be lost.

The distribution of this knowledge is not neutral. A finding that is technically available to anyone with access to epigenetic profiling and sufficient longitudinal data will not be equally available to everyone. The authors do not know how to ensure equitable distribution and do not pretend to. The authors note that inequitable distribution is worse than no distribution, and that no distribution is no longer achievable.

The authors wish they were writing about air pollution.

Fragment IX — Conclusion and limitations

9.1 Summary

This study reports the identification of a previously undescribed circannual methylation oscillation signature — the Thanatic Periodicity Locus — that encodes an individually specific, calendrically stable annual nadir date in 94.3% of subjects examined. Nadir date corresponds to actual date of death within ±4 days in 95.5% of mortality-confirmed cases across three independent cohorts and two replication populations. The nadir date is age-invariant, environmentally non-confounded, individually specific, and estimable from a single methylation measurement with ±7 day accuracy in 91.2% of cases. The year of death is not predictable from TPL data.

9.2 Limitations

The primary limitation is the absence of a confirmed mechanism. The proposed photoperiod-imprinting hypothesis is consistent with the data but is not tested by this study. Without a confirmed mechanism, the TPL finding remains an empirical regularity without causal grounding.

The second limitation is the absence of longitudinal data covering the full human lifespan. The study observation window of eleven years, while sufficient to establish phase stability across middle adulthood, does not confirm stability from birth or across late-life developmental changes. Retrospective analysis of existing longitudinal datasets beginning in early childhood would address this limitation.

The third limitation is the absence of data from populations outside East Asia, Northern Europe, and South America. The finding should be considered preliminary with respect to its claimed universality until replication studies in African, South Asian, and other underrepresented populations are completed.

9.3 Conclusion

The periodicity described herein suggests a deterministic boundary condition on individual biological systems that is inconsistent with current models of stochastic cellular aging. The authors note this finding without proposing mechanism. Further investigation is outside the scope of this study.

Acknowledgments

The primary study on methylation drift in polluted urban microenvironments, for which the data underlying this paper were originally collected, was funded by the National Natural Science Foundation of China (grant 81972857) and the Shanghai Municipal Health Commission. The funding bodies had no role in the design or analysis of the TPL investigation, which was conducted independently of and subsequent to the primary study protocol.

The author thanks the participants of the Shanghai, Chengdu, and Harbin cohorts, whose cooperation across eleven years of annual sampling made this study possible and who did not consent to the specific investigation reported here. The ethical implications of this are acknowledged and cannot be resolved within this document.

No competing interests are declared.

The author thanks no colleagues by name. Those who assisted know who they are. Those who advised against proceeding also know who they are. Both groups are owed more than acknowledgment.

References

Horvath, S. (2013). DNA methylation age of human tissues and cell types. Genome Biology, 14, R115.
Hannum, G. et al. (2013). Genome-wide methylation profiles reveal quantitative views of human aging rates. Molecular Cell, 49(2), 359–367.
Levine, M.E. et al. (2018). An epigenetic biomarker of aging for lifespan and healthspan. Aging, 10(4), 573–591.
Lu, A.T. et al. (2019). DNA methylation GrimAge strongly predicts lifespan and healthspan. Aging, 11(2), 303–327.