What's in a GenoSight Report: The 9 Health Domains

Why context changes the output

Two people can carry the exact same variant and need completely different actions. A C677T heterozygous result with a clean homocysteine lab and no symptoms is a watch-and-wait finding. The same variant in someone with elevated homocysteine, low B12, and treatment-resistant depression is the most actionable thing in the report.

Tools that don't capture context can't make that distinction. They produce the same paragraph for both people. (How that synthesis actually runs.) GenoSight's onboarding is built around the idea that the variant is half the data; the rest is who you are.

The nine domains

Each domain feeds into the synthesis prompt as structured context. None of them is a free-text dump — the onboarding chat asks targeted questions, and the answers get stored as typed fields the engine can reason about.

1. Demographics

What we ask: age, sex assigned at birth, ancestry (best estimate), height, and weight.

Why it matters: allele frequencies vary across populations, so "rare variant" needs a denominator. A SNP that's rare in European-ancestry cohorts may be common in East Asian cohorts, and the clinical interpretation shifts accordingly. Sex affects PGx for several drug classes (warfarin dosing, hormonal medications). Body composition affects dosing math for some interactions.

2. Diagnoses

What we ask: current and past medical conditions, with rough date of diagnosis where you have it.

Why it matters: the report prioritizes findings relevant to what you're already managing. If you have diagnosed depression, CYP2C19 status moves to the top of the report because it predicts SSRI response. If you have hypertension, ADD1 and AGTR1 sodium-sensitivity variants get highlighted. Without diagnoses, the engine has no way to know which of dozens of findings is the most consequential for you specifically.

3. Family history

What we ask: first-degree (parents, siblings, children) and second-degree (grandparents, aunts, uncles) diagnoses where known.

Why it matters: family history multiplies the weight of certain variants. A BRCA-adjacent variant with strong family history reads differently than the same variant in isolation. For carrier status, family history affects whether a heterozygous result is informative or actionable.

4. Symptoms

What we ask: current health concerns or symptoms you'd want the report to address.

Why it matters: symptoms guide the executive summary. If you mention chronic fatigue, the report looks specifically at variants relevant to mitochondrial function, B-vitamin metabolism, iron handling, and thyroid pathways. If you mention insomnia, the synthesis pulls in CLOCK, PER3, COMT, and ADORA2A findings.

5. Current stack

What we ask: medications and supplements you're currently taking, with dose and form where you have it.

Why it matters: this is the highest-leverage domain. A methylated B-complex finding reads completely differently for someone already taking 5-MTHF than for someone on synthetic folic acid. Stack-aware synthesis is what makes the report useful instead of theoretical.

6. Allergies

What we ask: drug allergies, food allergies, environmental sensitivities.

Why it matters: allergy information lets the report rule out specific suggestions (some methylated B-complexes are derived from corn, for example).

7. Habits

What we ask: typical diet pattern, sleep duration and quality, exercise frequency and type, alcohol intake, caffeine intake, smoking status.

Why it matters: lifestyle SNPs only translate into recommendations through the lens of current habits. A high-caffeine consumer with a CYP1A2 slow-metabolizer genotype gets a meaningfully different report than someone with the same genotype who already drinks decaf. ADRB2 exercise-response variants matter more if you're training; less if you're not.

8. Goals

What we ask: what you want to optimize for — performance, longevity, fertility planning, managing a specific condition, prepping for a clinical visit.

Why it matters: goals decide tone and depth. A "prepping for a clinical visit" goal triggers a clean PDF optimized for handoff to a clinician. A "longevity optimization" goal pulls in lifespan-associated findings that wouldn't otherwise be prioritized.

9. Recent labs

What we ask: numerical values from any recent blood work — lipid panel, CBC, comprehensive metabolic panel, hormones, inflammatory markers, vitamin levels, homocysteine.

Why it matters: labs are the closest thing to ground truth for many of the pathways the engines reason about. A variant suggesting impaired vitamin D conversion is a hypothesis. A vitamin D level of 18 ng/mL alongside that variant turns the hypothesis into an action item. Labs also let the report set realistic monitoring schedules — "recheck homocysteine in 12 weeks" is concrete only if there's a baseline.

A worked example

The same finding, two different reports:

Profile A. 34, female, no diagnoses, no medications, no symptoms, balanced diet, regular exercise, recent homocysteine 8 µmol/L. Carries MTHFR C677T (CT heterozygous).

The report: low-priority finding, brief mention. Synthetic folate is fine; methylated form unnecessary unless preferences differ. No retesting recommended.

Profile B. 47, female, diagnosed major depressive disorder (treatment-resistant), on sertraline, also taking generic B-complex with folic acid, recent homocysteine 14 µmol/L, fatigue and low energy listed in symptoms. Carries the same MTHFR C677T (CT heterozygous).

The report: high-priority finding, prominent placement. Switch to methylated B-complex with 5-MTHF and methylcobalamin; discuss CYP2C19 status with prescriber regarding sertraline metabolism; recheck homocysteine in 12 weeks. Bring printed PDF to next psychiatry appointment.

Same gene, same genotype. Different person. Different report.

What you don't have to provide

The onboarding is gated — every domain is optional except the bare minimum required for the engines to run. Reports get more useful as you fill in more, but you don't have to share anything you don't want to. You can update domains later and regenerate the relevant report sections.

Sensitive data is stored encrypted at rest with row-level security so each user can only read their own data. The raw genotype file is stored separately and is never sent to the LLM during synthesis — only the structured engine findings. (Full breakdown of how raw DNA data is handled.)

For the broader picture of how the report is produced — the parsers, the engines, and the synthesis pass — see what GenoSight is and how AI reads your DNA.

Key takeaways

• A GenoSight report reasons about your genetics against nine health-context domains, not in isolation.
• The nine domains: demographics, diagnoses, family history, symptoms, current stack, allergies, habits, goals, recent labs.
• Same variant + different context = different report. The MTHFR C677T worked example shows this concretely.
• The current-stack domain is the highest-leverage input — most recommendations only make sense relative to what you're already taking.
• Onboarding is gated; you only fill in what you want, and you can update domains later to refine the report.