~/shanegraffiti.com/research/genomic-anchor Shane Graffiti Inc. AI Research Division 2026

IS IT YOU OR YOUR ENVIRONMENT?

Personalized health AI faces a cold-start problem: models need weeks of behavioral data before they can tell constitutional variation from environmentally driven deviation. This framework proposes an exogenous genetic anchor fixed at conception, immune to reverse causation, available before a single reading is taken as the Bayesian prior that bridges the gap. The same observed HRV of 55 ms generates a suppression hypothesis for a person whose genomic prior predicts 80 ms, and an enhancement hypothesis for a person whose prior predicts 30 ms a reversal impossible without a personalized anchor.

Aruna Dey, Suraj Biswas
Authors
Dots-In (IIT Bombay incubated)
Affiliation
arXiv 2606.13556 Jun 2026
Published
Exogenous Genetic Anchor
Core Mechanism
Bayesian Priors◆ Genomic Anchoring◆ Causal Decomposition◆ Mendelian Randomization◆ Cold-Start Problem◆ Polygenic Scores◆ FTO / FADS1 / FKBP5◆ N-of-1 Causal Inference◆ Prior Decay◆ Ancestry-Matched Effect Sizes◆
§ 1.0

The Cold-Start Problem

A personal behavioral baseline is the most informative reference for physiological interpretation, but it takes roughly 7–30 days of consistent data per signal to stabilize. Before that, a system running on population norms cannot tell a constitutionally high signal from an environmentally elevated one. This is not a data-quantity problem it is a reference-quality problem. Population norms are the wrong reference for an individual, and no amount of additional data collection fixes that.

Reference 1 Population Norms 01

Group-Level
Typicality

Answers whether a reading is typical for the population. Cannot say whether it is typical for this individual a constitutionally high-tone person looks "normal" on every table even when severely suppressed, because the suppressed value still falls in range.

⇄ bridged by
Reference 2 Genomic Anchor 02

Day-Zero
Personalization

Exogenous because genotype is fixed at conception and cannot be caused by downstream behavior, environment, or state available before a single behavioral observation, bridging the gap until a personal baseline stabilizes.

The genetic set point Ĝ = μ + Σᵢ βᵢ gᵢ // μ = population mean; βᵢ = GWAS-derived effect size; gᵢ ∈ {0,1,2} risk-allele count δ = P − Ĝ ~ N(P − μG, σ²G + σ²ε) // the deviation is the candidate-causal, actionable signal
§ 3.4

The "Normal for Whom" Reversal

Two people show an identical observed HRV of 55 ms both fall inside the population norm band. Against their genetic set points, the same number means opposite things.

Person A High Set Point 01

δ = −25 ms

Genetic prior predicts 80 ms. Observed 55 ms sits well below expectation. Reading: physiological suppression sleep debt, training load, or chronic stress are the candidate causes.

Person B Low Set Point 02

δ = +25 ms

Genetic prior predicts 30 ms. Observed 55 ms sits well above expectation. Reading: environmental support current conditions are favoring autonomic function beyond constitutional baseline.

§ 4.0

Six Physiological Domains

The decomposition's value depends entirely on the quality of Ĝ, which varies sharply by domain. Strongest anchors carry tight uncertainty bands; weak ones widen the band until larger deviations are required before any attribution is generated.

DomainKey gene(s) / set pointEvidence
Metabolic / AppetiteFTO (rs9939609) satiety threshold, resting metabolic rate tendency. The most replicated common variant for body mass in the human genome.Strong
Fatty-Acid / InflammatoryFADS1/2 constitutive PUFA ratio, systemic inflammatory tone. Two people on identical diets can show different inflammatory baselines genetically.Strong
Stress-Axis / CortisolFKBP5 (rs1360780) HPA feedback speed, cortisol-recovery ceiling. Interacts with early-life adversity through epigenetic demethylation.Moderate–Strong
Autonomic TonePolygenic (GNG11, RGS6, HCN4) resting HRV, resting heart rate. Real but spread across dozens of loci explaining only 0.9–2.6% of variance.Moderate, Polygenic
Circadian / ChronotypePolygenic (PER1/2/3, CRY1, ARNTL) constitutional sleep midpoint, circadian phase. 351 loci identified across 697,828 individuals.Moderate, Polygenic
Dopaminergic / SerotonergicCOMT, DRD2, SLC6A4 prefrontal dopamine tone, reward sensitivity, serotonin reuptake. Largely failed large-scale replication.Weak / Contested
§ 4.1

Worked Example FTO

A person carries two copies of the FTO risk allele (rs9939609 A/A). Without a genetic anchor, evening snacking and slow satiety responses generate a false behavioral attribution the system ranks poor habits or stress as the top causal candidates. With the exogenous genetic anchor, the deviation from the expected metabolic baseline is much smaller, and the ranked hypothesis correctly identifies a constitutional low-satiety signal amplified by an obesogenic food environment.

0.36 FTO β per risk allele, kg/m² BMI
0.30 FADS1 β, circulating PUFA ratio (SD)
0.19 FKBP5 β, cortisol AUC (SD)
0.04 COMT β, dopamine proxy (SD) weakest tier

Strong metabolic anchors carry 4–9× larger effect sizes than the dopaminergic candidate gene the gap between a real constitutional anchor and a contested one is not subtle.

§ 4.6

The Cautionary Tier

The genes most widely sold in consumer genomics panels are precisely the ones that have failed rigorous large-scale replication. Single-gene thinking applied to massively polygenic traits was always going to disappoint this table is the receipt.

Gene (variant)Claimed associationReplication status
COMT (Val158Met)Prefrontal dopamine, executive functionEnzyme effect robust; behavioral effects small & context-dependent
SLC6A4 (5-HTTLPR)S-allele raises anxiety/depression under stressLarge pre-registered replications found no robust effect
MAOA (uVNTR)"Warrior gene" impulsivity, aggressionSmall, inconsistent; documented history of forensic misuse
DRD2 (TaqIA)"Reward deficiency" addiction riskDensity effect debated; modest & contested in meta-analysis
DRD4 (7R VNTR)Novelty-seeking, ADHD riskMixed in meta-analyses; effect small
DRD3 (Ser9Gly)Altered D3 affinity, impulsivityWeak and inconsistent
§ 5.0

The Causal Ladder

The framework is explicit about what it can and cannot deliver. It climbs only the first rung from observation; stronger causal claims require intervention or counterfactual evidence the framework does not by itself supply.

{{ s.n }} {{ s.name }}
{{ s.tag }} {{ s.name }}
{{ row.key }} {{ row.val }}
§ 6.0

The Prior-Decay Architecture

Because the exogenous genetic anchor is weak, it should not persist at full weight once behavioral data exist. As longitudinal data accrue, the empirical personal baseline progressively replaces the genomic prior as the primary reference settling at a non-zero floor so the anchor keeps contributing for sparse signals and after long data gaps.

Dynamic belief update Ĝₜ = w(t) Ĝgenomic + [1 − w(t)] P̄ₜ // w(t) → 1 at cold-start (t = 0); w(t) → wmin > 0 as t → ∞ // floor ≈ 30%: the genetic anchor never fully disappears as an interpretive reference
§ 7.0

Four Constraints for Honest Deployment

Calibrated restraint is the framework's defining discipline. These four constraints exist to prevent a genuinely informative but genuinely weak prior from being used as if it were a verdict.

ConstraintRequirement
{{ t.name }} {{ t.body }}

NATURE OR NURTURE ISN'T THE QUESTION. THE DEVIATION IS.

Is It You or Your Environment? A Bayesian Inference Framework for Genomically-Anchored Personalized Physiological Interpretation.

Shane Graffiti Inc. AI Research Division Top ↑arXiv 2606.135562026