Ben Sima — Genetically-Derived Diet Framework

Based on 23andMe v4 genome analysis, March 2026

Evidence confidence tiers

Each recommendation below is tagged with an evidence confidence level:

• 🟢 HIGH — Hard mechanistic biology (single gene → known protein → known effect). Not dependent on nutrition epidemiology.
• 🟡 MEDIUM — Real gene-diet interaction studies (RCTs or strong cohort data), plausible mechanism, but limited replication or modest sample sizes.
• 🟠 LOW-MEDIUM — Subgroup analyses, post-hoc reanalyses, or plausible inference from stacking multiple modest-effect SNPs. Treat as hypothesis to test on yourself.
• 🔴 LOW — Reasonable inference but no direct study of this specific genotype-diet combination. Speculative.

Tier 1: Strong genetic constraints (modify diet around these)

1. Low saturated fat, emphasize omega-3 (APOC3 double-homozygous risk)

Confidence: 🟡 MEDIUM

Mechanism (HIGH confidence): APOC3 promoter variants → elevated apoC-III → inhibited lipoprotein lipase. This is well-established molecular biology.

Diet interaction (MEDIUM): The Framingham Offspring Study showed APOC3 risk carriers have amplified trig response to saturated fat. This is real cohort data with a plausible mechanism, but it’s not an RCT. The fish oil recommendation is supported by separate RCT evidence for EPA + trig lowering (REDUCE-IT trial), though that trial wasn’t stratified by APOC3 genotype.

Rules:

• Minimize saturated fat: butter, cheese, fatty red meat, coconut oil
• Primary fats: olive oil, avocado oil, fish oil
• Fish oil 2-4g EPA+DHA daily (EPA specifically upregulates LPL)
• Fatty fish 3-4x/week (salmon, sardines, mackerel)
• When eating meat: lean cuts, poultry, fish

2. Eat earlier in the day (MTNR1B heterozygous risk)

Confidence: 🟡 MEDIUM (timing) / 🟠 LOW-MEDIUM (low-fat composition)

Timing mechanism (HIGH): Melatonin suppresses insulin secretion via MT1B receptor on beta cells. G allele → receptor overexpression. This is well-established cell biology.

Timing clinical evidence (MEDIUM): One RCT (Garaulet et al. 2022, Diabetes Care) showed G carriers have worse glucose tolerance eating late vs. early. Real trial, but single study, modest N.

Low-fat diet composition (LOW-MEDIUM): From a subgroup reanalysis of the POUNDS Lost trial. Subgroup analyses of diet trials are exactly the kind of finding that often fails to replicate. This is the weakest claim in the document. Treat as speculative.

Rules:

• Primary meal before 5pm, ideally midday
• If evening eating: keep it early (before melatonin onset ~9pm)
• Bright light exposure through the meal
• Low-fat diet composition preferred over high-fat (speculative — based on subgroup analysis)

3. Dairy-free or dairy-minimal (LCT GG — lactose non-persistent)

Confidence: 🟢 HIGH

This is a single enzyme gene. GG = you don’t produce lactase as an adult. Binary, replicated thousands of times across populations. Not debatable as genetics. The magnitude of symptoms varies person-to-person, but the enzyme absence is certain.

Rules:

• Eliminate liquid milk, soft cheese, ice cream
• Hard aged cheeses (parmesan, aged cheddar) have minimal lactose — tolerable
• Fermented dairy (yogurt, kefir) partially pre-digested — test individually
• If using dairy: lactase enzyme supplement

Tier 2: Moderate genetic nudges

4. Anti-inflammatory emphasis (SH2B3 TT homozygous + AAT carrier)

Confidence: 🟡 MEDIUM (AAT carrier risk) / 🟠 LOW-MEDIUM (specific diet rules)

AAT PiMZ biology* (HIGH): Z protein polymerization in hepatocytes and ER stress is well-established molecular biology.

AAT + NAFLD interaction (MEDIUM): Several cohort studies show Pi*MZ carriers have 2-3x increased fibrosis risk when a second liver insult is present. Not an RCT, but consistent across studies.

SH2B3 inflammation (MEDIUM): GWAS-validated inflammatory variant. The specific diet recommendations (polyphenols, omega-3:6 ratio) are extrapolated from general anti-inflammatory diet literature, not SH2B3-specific studies.

Specific food rules (LOW-MEDIUM): “Eat anti-inflammatory foods” is reasonable but the specific items (berries, turmeric, etc.) have individually weak clinical evidence. The overall pattern is more important than any single food.

Rules:

• High polyphenol intake: berries, dark leafy greens, green tea, turmeric
• Omega-3:omega-6 ratio matters more for you than average
• Minimize seed oils (high omega-6)
• Cruciferous vegetables daily (also support GSTP1 heterozygous detox)
• Avoid alcohol (AAT carrier + ADH1B fast metabolizer = double reason) (alcohol avoidance for AAT carriers is HIGH confidence)

5. Preformed omega-3 only (FADS1 heterozygous)

Confidence: 🟡 MEDIUM

Mechanism (HIGH): FADS1 encodes delta-5 desaturase. Heterozygous = intermediate ALA→EPA conversion. Well-characterized enzyme.

Practical magnitude (MEDIUM): How much this matters day-to-day is debatable. You’d still convert some plant omega-3. But given you already need fish oil for APOC3, this is a free stacking recommendation — no behavioral cost.

Rules:

• Fish oil or fatty fish for omega-3 (not plant sources)
• This stacks with APOC3 recommendation above

6. Supplement vitamin D (VDR/GC variants)

Confidence: 🟡 MEDIUM

Genetics (MEDIUM): VDR and GC variants associated with lower 25(OH)D in GWAS. Effect sizes are modest.

Recommendation strength: Vitamin D supplementation for someone who is dairy-free, does OMAD, and has these variants is reasonable regardless of genetic confidence. The blood test (25(OH)D) will confirm whether you actually need it.

Rules:

• Vitamin D3 2000-4000 IU/day
• Take with fat for absorption
• Monitor 25(OH)D, target 40-60 ng/mL

7. Methylated B vitamins (TCN2 + MTRR + MTHFR A1298C het stack)

Confidence: 🟠 LOW-MEDIUM

Individual SNP biology (MEDIUM each): Each variant has a known effect on B12 transport or recycling. Well-studied individually.

Stacking inference (LOW-MEDIUM): The claim that these three heterozygous variants together create a “functional B12 underperformer” is my inference, not something a study showed for this specific combination. Each has a modest effect; whether they compound meaningfully is uncertain.

Validation available: MMA and homocysteine blood tests will answer this definitively. Don’t trust the genetics alone — test it.

Rules:

• Methylcobalamin (not cyanocobalamin)
• Methylfolate (not folic acid)
• B-complex with methylated forms
• Validate with MMA + homocysteine on April blood draw before committing to long-term supplementation

Tier 3: Genetic-diet synergies with your existing protocols

8. Your existing protocols are genetically well-matched

Confidence: 🟠 LOW-MEDIUM (as a synthesis)

This section stacks multiple individual findings into a narrative. Each bullet is only as strong as its underlying evidence. The overall direction is probably right, but the precision is overstated.

Simple carb refeeds (fruit, honey, rice):

• ✅ Low-fat = good for MTNR1B genotype (LOW-MEDIUM — subgroup analysis)
• ✅ Low saturated fat = good for APOC3 (MEDIUM)
• ✅ Dairy-free naturally = good for LCT (HIGH)
• ✅ Anti-inflammatory (fruit polyphenols) = good for SH2B3 (LOW-MEDIUM)

Extended fasting:

• ✅ Lowers triglycerides (compensates for APOC3) (MEDIUM)
• ✅ Reduces liver fat (protects AAT-vulnerable liver) (MEDIUM)
• ✅ Improves insulin sensitivity (compensates for missing PPARG protective allele) (LOW-MEDIUM)

Zone 2 exercise on refeed days:

• ✅ BDNF Val/Val = full neuroplasticity benefit from exercise (HIGH for genotype, LOW-MEDIUM for practical magnitude)
• ✅ FTO heterozygous = exercise mitigates appetite effect (MEDIUM)
• ✅ Glycogen depletion → GLUT4-window eating = your metabolic throughput theory is genetically sound (LOW-MEDIUM — plausible but untested as a specific protocol)

9. The ideal day (synthesis)

Confidence: 🟠 LOW-MEDIUM overall — this is an optimized guess, not a prescription

Fasting day: Water, electrolytes, green tea. Exercise optional (liver fat oxidation is the goal).

Eating day:

• Morning: Zone 2 exercise (deplete ~50-70g glycogen)
• Midday meal (11am-2pm): Post-workout in GLUT4 window
  • Lean protein (chicken, fish, egg whites)
  • Simple carbs matched to glycogen depletion (fruit, rice, honey)
  • Minimal added fat (olive oil only, small amounts)
  • No dairy
  • Dark leafy greens, cruciferous veg
• If second meal needed: before 5pm, same composition
• Supplements with meal: fish oil, vitamin D3, methylated B-complex

What NOT to eat (genetic contraindications)

Genetic non-issues (things you DON’T need to worry about)

Confidence: 🟢 HIGH (absence of risk alleles is straightforward)

• Gluten: No HLA-DQ2.5 celiac risk. Gluten is fine.
• Iron overload: No HFE mutations. Don’t restrict iron-rich foods.
• Caffeine: COMT heterozygous = moderate metabolizer. Coffee is fine.
• Folate from food: MTHFR is normal. Eat folate-rich foods normally (supplement methylfolate as insurance for B12 recycling stack).