Mariam Hourani — Neotrium Bio-Architecture

Start here · what this is built on

Six of the twelve files were never opened.

Read this page first. It tells you which parts of the previous workbook to trust — and which were written without evidence that was already sitting in the folder.

Six source documents were never read when the workbook was built. Four of them are blood results that answer the exact questions it lists as unresolved.

679,568 SNPs · 12 files · 2 labs · 4 blood panels · ~500 genetic claims

Raw genotype

679,568SNPs

Verified · GRCh38 · forward strand

Blood draws

Nov 2025 → Apr 2026

Files never read

Four of them lab panels

Genetic calls wrong

Of ~120 that can be checked

The transcription was excellent. The coverage was not.

What was checked	Result	Meaning
Weight & Body Fat tabs — values vs source	● 233 / 233 match	Nothing was mistyped
CircleDNA marker rows	● 328 / 328 match	Nothing was mistyped
Fitness tab	● 37 / 38 match	One paraphrase, same direction
Source files actually opened	● 3 of 12	The gap is coverage, not accuracy
Blood results incorporated	● 0 of 4 panels	Three top priorities were already answered

A workbook can be perfectly accurate about the wrong third of the evidence. That is what happened here.

⚠ Act on this

Ferritin 10.80 was in the folder, unread.

Iron deficiency without anaemia, measured 27 Apr 2026. It appears nowhere in the workbook.

It is the only new, measurable, correctable abnormality in twelve files.

⚠ False alarm

The "98th percentile" was already disproven.

Metabolic syndrome was ranked priority #1. Her Nov 2025 bloods meet 0 of the 5 diagnostic criteria.

⚠ Wrong premise

The "missing" file isn't missing.

The workbook states the 267-page Diet & Nutrition panel "has never come through." It has been in the folder since 29 Jul 2025.

Order of evidence

Blood beats genotype. Genotype beats commentary.

A measured ferritin outranks a predicted iron need. A raw genotype outranks a PDF's label. Both outrank a summary sheet.

The data behind this chapter.

Every file in the folder, what it is, and whether the workbook ever registered it.

12 FILES

08 Nov 2025: Dio Life, 12-page panel — fasting glucose, HbA1c, full lipids, LFTs, iron studies, B12, 25-OH vitamin D. The panel that closes the metabolic-syndrome question.
27 Apr 2026: Mediclinic — ferritin, full blood count, tissue transglutaminase IgA and IgG. One draw, one lab number, one morning. Someone was already investigating iron properly.
Genotype: Gene Origins / NutriGenix array, 18 Jul 2025 — 679,568 markers, GRCh38, forward strand. The only raw data either lab released.
CircleDNA: Whole-exome, sample 2710-2173-3025-11, report 23 Dec 2025, 85 pages. Zero rsIDs and zero genotypes printed. Nothing in it can be checked from the customer's side.
Gut microbiome: 43-page stool report. Pages 18, 19, 22 and 23 encode their results in colour only — unreadable from the text layer, in either direction.

What was in the folder, and what the workbook used

File	Size	Used?	What it holds
CircleDNA Premium	85 pp	Used	~200 trait calls, 35 cancers, 103 carrier conditions, drug panel pp. 55–61
Fitness Summary	73 pp	Used	Endurance, recovery, ACL (39 variants, not 21)
Weight & Body Fat	83 pp	Used	PLIN, AMY1A, TFAP2B, MTNR1B
Diet & Nutrition	267 pp	Never opened	117 genetic results — the only panel that cross-checks the whole Nutrition tab
Food Sensitivities	60 pp	Never opened	21 results, 15 on axes CircleDNA never tested — oxalate, histamine, casein, milk/egg/peanut/shrimp
Gut Microbiome	43 pp	Never opened	11 keystone species low, Prevotella copri 36%, 8 vitamins flagged
Dio Life panel	12 pp	Never opened	Closes metabolic syndrome, NAFLD screening and vitamin D
Mediclinic × 3	—	Never opened	Ferritin 10.80 · coeliac serology negative
Clinic doctor's sheet	4 pp	Never opened	See Act VII — it contradicts the report it claims to summarise
Raw genotype TXT	679,568 SNPs	Never used	The arbiter for every gene-label dispute the workbook flagged

Bonus finding from the raw file: the array carries 679,568 genotyped markers. The panels claim to have "looked at" 816,639 (oxalate), 728,984 (belly fat) and 636,870 (metabolic syndrome) variants — more than were ever measured. Those are imputed polygenic scores printed as though they were measured counts.

Sex check: 0 chrY markers · 24.6% chrX heterozygosity → female ✓
Mitochondrial: Correctly haploid ✓
Strand: Confirmed forward at four independent markers ✓
Independence: The array (Jul 2025, NutriGenix) and CircleDNA (Dec 2025, exome) are different samples on different platforms. Germline genotype does not change between July and December — so this is genuine cross-validation, not a self-check.
Trap worth recording: Duplicate probes are named with a numeric suffix (rs12248560.1, rs9923231.1). Naive matching returns "not on array" for two of the most decisive variants in the file. Both are present, and both overturn a CircleDNA call.
Call rate: 679,568 of 679,568 — zero missing calls. Real arrays run 98–99.8%. No-calls were either dropped before export or filled with reference; either is defensible, but it isn't disclosed, so measured cannot be told from assumed.

Act I

Of ~500 claims, one needs a doctor.

Ferritin 10.80 against a floor of 30. Measured, not predicted. Correctable.

Chapter 01 · Iron

Your stores are empty. Your blood hasn't noticed yet.

Iron deficiency without anaemia — the earliest stage, caught before haemoglobin moved. The genotype explains why you're vulnerable and confirms it's safe to treat.

10.8ng/mL

Reference floor is 30. Haemoglobin 13.8, MCV 89.4 and MCH 30.0 are all comfortably normal — this was caught before anaemia.

Ferritin 10.80 · 27 Apr 2026 · Mediclinic · coeliac already excluded

Iron status · where each marker sits against its range

Ferritin — stored iron10.80 ng/mL Low · ref 30–120

Haemoglobin13.8 g/dL Normal

MCV — red cell size89.4 fL Normal

Transferrin saturation Nov 2528 % Normal — lower half

TIBC Nov 25387 µg/dL Normal — upper half

CRP — is the ferritin real?not tested Missing

Green band = reference range. TIBC in the upper half and saturation in the lower half is the signature of low stores — that pattern was already visible in November, five months before anyone measured ferritin.

Nov 2025 → Apr 2026 · the only marker that moved

RDW — red cell variation 13.3% → 14.0%

Haemoglobin 13.4 → 13.8

Ferritin never measured → 10.80

An important caveat on the timeline. Ferritin was never measured in November and iron studies were never repeated in April. The two panels don't overlap, so we cannot say the ferritin "dropped" — it may simply never have been looked for. RDW nudging upward is the first thing to move in iron deficiency, and it moved.

What the genes say

A genetically poor absorber — with no overload risk.

Heterozygous across the entire TMPRSS6 block plus TF. Higher hepcidin, so less dietary iron gets absorbed and stores replenish slowly. All three HFE variants wild-type — no haemochromatosis, so repletion carries no hazard.

TMPRSS6 rs855791 AG · TF rs3811647 AG · HFE all WT

What the blood says

Stores depleted, production intact.

Ferritin 10.80 with normal Hb, MCV, MCH and RDW. TF rs3811647 predicts raised TIBC and lowered saturation — and her measured TIBC is 387 with saturation 28%. The prediction and the measurement match.

Mediclinic 27-Apr-2026 · Dio Life 08-Nov-2025

The genotype and the bloods tell the same story, independently. That is rare in this folder, and it is why this finding is solid while most of the rest is not.

Three recommendations in her own reports work directly against this

"Avoid vitamin C supplements — permanently" (Food Sensitivities rec #5, oxalate rationale). Vitamin C is the main enhancer of non-haem iron absorption. And the oxalate call behind it is the weakest in the file — a whole-array score with no percentile given.

"Calcium 1000 mg daily" (rec #10 + the panel's "increased calcium need"). Calcium inhibits iron absorption when taken at the same time. Not wrong to supplement — wrong to co-time.

"1–3 cups of black coffee daily" (Weight report rec #41). Coffee and tea polyphenols substantially reduce non-haem iron absorption when taken with or near meals.

Layer the dairy restriction on top and she has been steered toward a pattern that blocks iron from several directions at once — while her measured ferritin was 10.8. None of the three needs abandoning. They need re-timing.

Five things to raise at one appointment

1 — Repeat, complete, one draw

Ferritin + iron + TIBC + saturation + CBC + CRPthe April number is now three months old

These four iron markers have never been measured together. CRP matters because ferritin rises with inflammation — a normal CRP confirms 10.8 is a true reading and not a falsely reassuring one.

2 — Find the cause

Menstrual history firstthen GI if unexplained

In a 41-year-old woman the differential is short: menstrual loss, GI loss, malabsorption, intake. Coeliac is already off the list. The deficiency is confirmed; the cause is not. That is the whole remaining task.

3 — Two genotype facts

No HFE variants · TMPRSS6 heterozygouschanges the practicalities

No overload risk, so repletion is not hazardous. Hepcidin runs high, so oral absorption is blunted — current evidence favours alternate-day rather than daily dosing precisely because it lets hepcidin fall between doses. The regimen is the doctor's call.

4 — Fix the timing conflicts

Separate calcium from ironcoffee away from meals

And don't let an unsupported oxalate recommendation remove vitamin C while she is iron deficient.

5 — The likely knock-on

Hair sheddingferritin, not genotype

CircleDNA flagged "Androgenetic Alopecia: Elevated." That call is not valid in women — the male pattern-baldness loci don't replicate in female pattern hair loss (Redler 2012). A ferritin of 10.8 is a far better explanation, and unlike a genotype it's correctable.

The data behind this chapter.

Every value, with its date, reference range and the genotype behind it.

4 SOURCES

Date	Test	Result	Reference	Flag
27 Apr 2026	Ferritin	10.80 ng/mL	30.00–120.00	LOW
27 Apr 2026	Haemoglobin	13.8 g/dL	11.5–16.0	normal
27 Apr 2026	MCV	89.4 fL	80–100	normal
27 Apr 2026	MCH	30.0 pg	24.7–32.8	normal
27 Apr 2026	RDW	14.0 %	12.3–17.7	normal
08 Nov 2025	Serum iron	108 µg/dL	51–150	normal
08 Nov 2025	TIBC	387 µg/dL	250–450	normal, upper half
08 Nov 2025	Transferrin saturation	28 %	15–50	normal, lower half
08 Nov 2025	Haemoglobin	13.4 g/dL	12–15	normal
08 Nov 2025	RDW-CV	13.3 %	11.6–14	normal

Gene	Variant	Genotype	Meaning
HFE	rs1800562 (C282Y)	GG	Not a carrier
HFE	rs1799945 (H63D)	CC	Not a carrier
HFE	rs1800730 (S65C)	AA	Not a carrier
TMPRSS6	rs855791 (A736V)	AG	Heterozygous for the low-iron allele — raises hepcidin, so less dietary iron is absorbed. Strongest common iron-status variant known.
TMPRSS6	rs4820268	AG	Same direction
TMPRSS6	rs2413450	TC	Same direction
TF	rs3811647	AG	Raises transferrin/TIBC, lowers saturation — matches her measured TIBC 387 and saturation 28%

This does not cause deficiency on its own. It means she has less reserve, replenishes more slowly, and is likelier to become deficient when something else tips the balance.

Coeliac disease: Excluded. Tissue transglutaminase IgA <0.50 and IgG <0.8, both negative (27 Apr 2026), and she carries no DQ2.5 (rs2187668 CC). Coeliac is one of the leading causes of unexplained iron deficiency in adults — so this matters.
Note on that draw: The ferritin, the full blood count and the coeliac serology were all one blood draw, same lab number, same morning. Someone was already investigating this properly and checking the right cause first. This isn't a finding that has been ignored — it's one that needs following up.
B12: 586 pg/mL (183–822) — normal, not contributing
Vitamin D: 71.37 ng/mL — sufficient, not contributing
Still open: Menstrual blood loss · GI blood loss · H. pylori and gastritis · dietary intake given the restrictions she's been placed under

CircleDNA: "Iron: Normal Needs." Wrong.
Gene Origins: "Increased iron need, ≥18 mg/day." Right direction — but for the wrong reason, and it contradicted itself. Page 100 calls for increased iron; page 188 calls the same TMPRSS6 genotype "typical activity."
So the evidence is: The raw genotype and the ferritin. Not either PDF.
Why it matters: This is the one place in the entire folder where a genetic panel beat the other on a measurable clinical endpoint — and it still couldn't explain its own reasoning.

Act II

Four red flags the bloods close.

Including the one the workbook called "the single biggest new finding."

Chapter 02 · Metabolic syndrome

The 98th percentile was a false alarm.

It drove the entire priority list. Her bloods meet none of the five diagnostic criteria — and the panel that flagged it claimed to analyse more variants than her array ever measured.

0of 5

Metabolic syndrome requires three of five criteria. She meets zero. The "98th percentile" score was an imputed polygenic estimate; the bloods are a measurement.

Glucose 89 · HbA1c 5.0% · TG 40 · HDL 69 · TC/HDL 2.7 · 08 Nov 2025

The five criteria, measured

Fasting glucose89 mg/dL Criterion not met · cut-off 100

HbA1c5.0 % Excellent

Triglycerides40 mg/dL Criterion not met · cut-off 150

HDL69 mg/dL Criterion not met · floor 50 for women

Total cholesterol : HDL2.7 Well inside optimal

Waist circumferencenot measured Needs a tape measure

Four criteria measured and clear, one never taken. Even if waist were positive, one of five is not metabolic syndrome. CircleDNA rated all five components average and was right; the 98th-percentile score was not.

What the false alarm cost

This single line drove the workbook's whole priority list. Meanwhile a ferritin of 10.80 sat in the same folder, unread. The alarm was on the wrong number.

The score also rests on a claim of 636,870 variants analysed for metabolic syndrome — on an array that measured 679,568 markers in total, of which 40–60% of the rsIDs the panels cite aren't on it at all. Those are imputed values printed identically to measured ones.

NAFLD flag · liver

The whole liver panel is normal.

AST 19 · ALT 14 · GGT 11 · bilirubin 0.7 · albumin 3.8 · ALP 54.

NAFLD can exist with normal enzymes, so the genotype flag isn't disproven — but the screening the workbook asks for has effectively been done.

Vitamin D · "higher needs"

71.37 ng/mL — near the top of range.

Sufficiency is 30–100. The genotype call was sound (risk alleles at all four loci) — it just never mattered. The stress-fracture pairing built on it needs updating too.

Gluten · "worse HLA-DQ genetics"

Coeliac excluded twice over.

tTG IgA <0.50 and IgG <0.8, both negative. Raw genotype rs2187668 CC — no DQ2.5.

Serologically and genetically ruled out. Do not carry the label forward.

The data behind this chapter.

The Dio Life and Mediclinic panels in full, including the minor abnormals nobody mentioned.

2 PANELS

The workbook says	The blood says	Verdict
Metabolic Syndrome 98th percentile — "priority #1"	Glucose 89 · HbA1c 5.0 · TG 40 · HDL 69 · non-HDL 118	RESOLVED — 0 of 5 criteria
NAFLD — elevated risk, "discuss liver screening"	AST 19 · ALT 14 · GGT 11 · bilirubin 0.7 · ALP 54	No biochemical liver injury
Vitamin D higher needs + stress fracture risk	25-OH vitamin D 71.37 ng/mL · calcium 8.6	RESOLVED — comfortably sufficient
Vitamin B12 — normal needs	B12 586 pg/mL (183–822)	Confirmed
HLA-DQ "worse genetics" / coeliac section	tTG IgA <0.50 · IgG <0.8 · rs2187668 CC	Excluded, both ways
Iron — normal needs	Ferritin 10.80 — LOW	THE MISS · see Act I

Uric acid 2.4: Low. Usually benign in a woman with this diet pattern; worth a mention, not a workup.
BUN / creatinine 25: High — the classic hydration signature. Dubai.
MPV 12.8: High. Isolated MPV elevation with a normal platelet count is rarely meaningful.
Lymphocytes 49.8%: High as a percentage; absolute count is what matters and the CBC is otherwise unremarkable.
Eosinophils 0.46%: Low. Not clinically significant on its own.
Why they're listed: None of these are alarming. They are here because a report claiming to be complete should contain them.

Act III

Two labs. Same person. Different answers.

Twenty-two traits where both called it and an objective answer exists.

Chapter 03 · Head to head

Gene Origins wins 9–5. Neither should be used.

Adjudicated against the raw genotype and, where available, her blood results. The margin is narrower than it looks, because CircleDNA's biggest win is bigger than any of Gene Origins'.

9–5

Gene Origins was right more often. It was still wrong on gluten, wrong on spice, and self-contradictory on iron inside a single document.

22 head-to-head traits · 4 both right · 1 split · 1 both wrong · 1 unresolved

Gene Origins

9wins

Carbs, taste, obesity, iron, 4 nutrients

CircleDNA

5wins

Incl. the metabolic-syndrome call

Both right

Lactose, caffeine, ACTN3, alcohol

Raw data released

1of 2

Only Gene Origins

Every trait where both labs called it and an objective answer exists

Trait	CircleDNA	Gene Origins	Objective answer	Winner
Lactose	Intolerant	Intolerant	rs4988235 GG + Arabic site rs41380347 AA	Both right
Caffeine	Higher	Higher	ADORA2A rs5751876 TT · CYP1A2 AC	Both right
Power / ACTN3	Low	Lower activity	577XX (proxies r²>0.98)	Both right
Alcohol	Normal	Typical	ALDH2 GG · ADH1B CC	Both right
Carbohydrates	Normal	Better response	TCF7L2 rs7903146 CC — favourable homozygote	Gene Origins
Taste	Super-taster	Intermediate	Het at all 3 TAS2R38 sites = medium taster	Gene Origins
Obesity	Elevated	Less likely	Zero risk alleles at FTO ×4 and MC4R ×2	Gene Origins
Weight regain	Elevated	Typical	Same — zero FTO/MC4R risk	Gene Origins
Iron	Normal need	Increased need	TMPRSS6 het ×3 · ferritin 10.8	Gene Origins
Vitamin E	Higher need	Typical	CYP4F2 TT → higher circulating vitamin E	Gene Origins
Selenium	Higher need	Typical	GPX1 rs1050450 GG — high activity	Gene Origins
CoQ10	Higher need	Typical	NQO1 rs1800566 GG — full activity	Gene Origins
Fat sensitivity	Normal	Saturated fat worse	APOA2 rs5082 GG · PPARG rs1801282 CG	Gene Origins
Vitamin D	Higher need	Typical	Risk alleles at all 4 loci — but measured 71.4	Split
Vitamin A	Higher need	Typical	BCO1 rs7501331 TT — reduced conversion	CircleDNA
Metabolic syndrome	All 5 average	98th percentile	HbA1c 5.0 · glucose 89 · TG 40 · HDL 69	CircleDNA
Recovery	Lower	Faster	ACTN3 XX → greater damage, slower recovery	CircleDNA
Endurance	High	Typical	PPARGC1A rs8192678 CC + ACTN3 XX	CircleDNA
ACL rupture	Higher	Typical	Lacks protective COL5A1 rs12722 CC (she is TC)	CircleDNA (weakly)
Gluten / HLA-DQ	not tested	"Worse genetics"	rs2187668 CC · tTG both negative	GO wrong, unopposed
Spice / chilli	Higher	Higher	1 of 3 risk genotypes; best 2 SNPs not on array	Both unsupported
Belly fat	Waist average	89th percentile	No tape measure on file	Unresolved

CircleDNA's biggest win is bigger than any of Gene Origins'. A false metabolic-syndrome alarm at the 98th percentile costs more than five small nutrient-need errors. Gene Origins' biggest win is the iron call — the one place a panel beat the other on a measurable clinical endpoint.

The asymmetry that matters more than the score

Only one of these labs gave her the raw data. CircleDNA prints no rsIDs, no genotypes, no star alleles, no variant list — nothing that can be checked. Everything in this audit was possible only because Gene Origins handed over the file.

If she had bought CircleDNA alone, none of its errors would be discoverable — by her, or by any doctor. That is not a point of etiquette. It is the difference between a test result and a claim.

Why CircleDNA fails the way it does · exome vs. array

Trait	Dominant variant	Where it sits	Visible to whole-exome?
Obesity	FTO rs1421085	intronic	✗
Obesity	MC4R rs17782313	188 kb intergenic	✗
Heart disease	9p21 rs10757278	intergenic lncRNA	✗
Type 2 diabetes	TCF7L2 rs7903146	intronic	✗
Atrial fibrillation	PITX2 rs2200733	intergenic	✗
Migraine	PRDM16, TRPM8, LRP1	intronic / intergenic	✗
Alopecia	20p11	intergenic	✗
NAFLD	PNPLA3 I148M	missense	✓

This predicts the exact error pattern observed. Coding variant → CircleDNA is right (NAFLD, vitamin A, the CYP pharmacogenes). Non-coding → wrong or blind (obesity, migraine, 9p21, chronotype). It inverts its own product: polygenic framing applied where the platform has least data, while its genuine strength — rare coding variant detection — is buried under the least informative label available, "Average Risk."

Being auditable is not the same as being flawless.

Four real defects in the Gene Origins raw file, and one correction to an earlier claim.

4 DEFECTS

Perfect call rate: 679,568 of 679,568. Real arrays run 98–99.8%. Measured cannot be distinguished from assumed.
Indels are broken: Zero indel genotype codes across all rows — every insertion/deletion coerced into a two-letter SNP genotype.
rsID annotation errors: At least two rsIDs assigned to two different chromosomal positions with different genotypes (rs137853280, rs397509164) — and the second sits in the BRCA1 region.
Heavy undisclosed imputation: 40–60% of the rsIDs the four PDFs cite are not on the array at all — 803 of 1,333 in Diet & Nutrition, 414 of 694 in Fitness. That is how a 679,568-marker array produces claims of "816,639 variants analysed." Imputed calls are printed identically to measured ones — including ACTN3 rs1815739, presented as a headline result.
And the interpretation layer: Contradicts itself inside a single document: "increased iron need" on p100 vs "typical TMPRSS6 activity" on p188. The Weight report's rec #10 cites TFAP2B as its rationale while its own TFAP2B page says the opposite.

The 99.9% concordance figure means less than it appeared to

The four PDFs were previously reported as 99.9% concordant with the raw TXT, framed as strong validation. Now that we know the raw file came from the same test, that number proves only that SelfDecode transcribes Gene Origins' array faithfully. It says nothing about whether the array's base calls are correct. The genuinely independent comparison is the head-to-head above: two samples, two labs, one person.

Who did what: Gene Origins sold the service · a NutriGenix-branded array did the genotyping · SelfDecode generated the four PDFs.
Why that matters: The lab work and the interpretation come from different companies. It is part of why the genotypes are clean and the conclusions are not.

What was found: Three heterozygous calls at ClinVar Pathogenic sites: KCNH2 (rs794728448), PALB2 (rs587780206), JAG1 (rs863223664).
Assessment: All three are false positives, with high confidence. All three are indels or multi-nucleotide variants — and the file contains zero indel genotype codes anywhere across 679,150 calls. JAG1 causes Alagille syndrome, a dominant childhood-apparent condition; she is 41 and well.
Base rate: Published false-positive rates for rare pathogenic variants on consumer platforms run 40–85% (Tandy-Connor, Genet Med 2018).
Do: Mention to a doctor only if there is relevant family history — sudden cardiac death or unexplained syncope (KCNH2), or early-onset breast/pancreatic cancer (PALB2). Otherwise leave them.
The lesson: Raw consumer genotype files should not be self-interpreted for rare pathogenic variants. And a report that says "Long QT Syndrome: Average Risk" with no variant list gives the customer no way to know any of this exists.

From Gene Origins: The raw genotype file. That is the product. The four PDFs are commentary on it, and the commentary is unreliable.
From CircleDNA: The cancer and carrier screening — 35 cancer types and 103 recessive conditions, all clear. That is what exome sequencing is genuinely good at, and it is the one section that plays to its strength.
Everything else: Better answered by her raw file and her blood results, which together cost nothing more than she has already spent.
If she tests again: She doesn't need another genetic panel. She has 679,568 markers on file and they don't change. What she needs is a repeat ferritin, and her NAT2 status and untested Factor V Leiden written into her medical record.
One undisclosed limitation: She is ~48% Middle Eastern / ~43% European. Polygenic scores are derived overwhelmingly from European cohorts and lose calibration in admixed individuals, regressing estimates toward the mean — a plausible contributor to why 79 of 83 disease calls came back "Average." A report making 83 individual-level risk statements should disclose this. It does not.

Act IV

Wrong about as often as right.

Thirty correct, fifteen wrong, thirty-eight resting on nothing at all.

Chapter 04 · Where the labels break

You are not obese-predisposed. Or a super-taster.

Three separate weight calls, all backwards, all from the same underlying error — and she has zero risk alleles across six concordant SNPs at the two largest-effect obesity loci in the genome.

0risk alleles

FTO ×4 and MC4R ×2, all non-risk. Her net polygenic obesity burden is roughly 0.25 kg/m² below the population mean — and the report called it "Elevated."

FTO TT·TT·TT·TT · MC4R TT·GG · SEC16B AA · FAIM2 GG — six SNPs, one direction

~120 checkable trait calls · how they land

About 34 further disease calls are "correct" only because they restate the population base rate — they would be equally right if generated without a DNA sample at all.

Obesity · weight regain · thinness

Three calls, all backwards, one error.

FTO rs1421085 TT, rs17817449 TT, rs9939609 TT, rs1558902 TT; MC4R rs17782313 TT, rs12970134 GG; SEC16B AA; FAIM2 GG.

Not one risk allele. The exome cannot see any of these — they are intronic and intergenic.

Taste · TAS2R38

Medium taster, not super-taster.

rs713598 CG · rs1726866 AG · rs10246939 TC — heterozygous at all three diplotype-defining sites = PAV/AVI. Super-taster requires PAV/PAV.

Matters because super-taster status is routinely used to excuse vegetable aversion.

Stress · COMT rs4680

Neither warrior nor worrier.

Warrior = GG. Worrier = AA. She is AG — the heterozygote, the single most common genotype, about half of all people.

The label is defined by exactly one SNP and she sits in the middle of it.

Sleep · self-contradiction

"Deep sleeper" and "insomniac" from the same gene.

MEIS1 rs113851554 GG — no risk allele. The ADA variant behind "Deep Sleeper" predicts better sleep quality. Two rows apart, opposite conclusions.

Chronotype

Nothing supports "morning lark."

CLOCK rs1801260 AG — she carries one copy of the eveningness allele. No PER2 advanced-phase variant. The actual morningness marker (PER3 VNTR) is an indel and can't be typed on an array.

Addiction calls

Both on the wrong side.

Smoking "less likely": CHRNA5 rs16969968 AG — carrier of the best-established addiction variant in the genome.

Alcohol "less likely": ADH1B CC, ALDH2 GG — she carries neither protective variant. Baseline, not below it.

Three nutrients called "higher need" at loci where she has the favourable genotype

Vitamin E — CYP4F2 rs2108622 TT reduces tocopherol catabolism, so circulating α-tocopherol runs higher. Direction reversed.

CoQ10 — NQO1 rs1800566 GG, full activity. NQO1 is the enzyme that reduces ubiquinone to ubiquinol; it is the entire mechanistic basis for a CoQ10 claim.

Selenium — GPX1 rs1050450 GG (Pro198), the high-activity genotype.

Only vitamin A survives: BCO1 rs7501331 TT, homozygous for reduced β-carotene→retinol conversion. That call is correct — and vitamin A is the one nutrient here where over-supplementing carries real risk, so it is worth getting right in both directions.

The rest of the wrong calls — and the 46% that was never checkable.

Including three traits the report grades "EXCELLENT" on evidence that does not exist.

3 LAYERS

Call	Her genotype	Reality
Strength profile: HIGH	ACTN3 577XX	XX means lower maximal strength and lower power — the same gene they used correctly to call "Power: LOW." One gene cannot give both answers.
Fatigue resistance: BELOW AVERAGE	AMPD1 rs17602729 GG	Wild-type homozygous — the favourable genotype at the only fatigue locus with real functional evidence.
Migraine: ELEVATED	MEF2D AA · FHL5 AA · MTHFR 677CC	Zero risk alleles at every replicated locus. TRPM8 TT is the ~80% majority genotype with near-zero discriminative value.
Androgenetic alopecia: ELEVATED	AR/EDA2R + 20p11	Scored on the male-derived scale. The male AGA loci do not replicate in female pattern hair loss (Redler 2012). Her ferritin of 10.8 is a far better explanation — and it's correctable.
Thrill-seeker	DRD4 VNTR · DRD2/ANKK1 rs1800497 GG	Built on a marker no SNP array can read. At the one typeable major locus she is the opposite of the claimed profile.
Skin hydration: NORMAL	FLG rs61816761, rs2065955	Both not on array. Filaggrin loss-of-function is the determinant of skin dryness. A definite result was printed for a trait with no data behind it.
Sunburn LOW + photoaging HIGH	TYR rs1126809 AA · ASIP het ×2 · MC1R R151C not typed	Mutually incompatible — UV sensitivity drives both — and "LOW" isn't supported by her pigmentation panel.
Photic sneeze: LESS LIKELY	rs10427255 TC	Carrier of the risk allele at the primary locus. Wrong side.
Body odour: MORE THAN NORMAL	ABCC11 rs17822931 CC	CC only separates "has odour" from "essentially none" — and ~98% of people of her ancestry are CC. A near-universal genotype dressed up as a personal finding.

Twenty-four of the 52 skin / trait / personality calls rest on no validated genetic evidence

All five Big Five traits · IQ · EQ · Creativity · Entrepreneurship · Educational Attainment · Information Processing · Language · Maths · Memory · Dancing Ability · Musical Ability · Altruism · Washing/Cleaning Obsessions · Smell Sensitivity · Skin Age · Skin Detoxification · Skin Lightening · Cellulite.

Entrepreneurship — "EXCELLENT": The largest GWAS of entrepreneurship (van der Loos 2013, n≈50,000, 16 cohorts) found zero genome-wide significant associations.
EQ — "EXCELLENT": The Warrier 2018 GWAS of the Reading-the-Mind-in-the-Eyes task found no genome-wide significant loci in its primary analysis.
Memory — "EXCELLENT": Rests on KIBRA rs17070145 — one of the most prominent failed replications in cognitive genetics.
The defect: Not a wrong answer. Printing a confident answer where the science supports none.
Plus 8 category errors: HCM, DCM, ARVC, Brugada, Long QT, Short QT, CPVT and "Familial Hypercholesterolemia: Negative" are Mendelian — risk is near-binary. "Average Risk" implies a graded polygenic scale that does not exist for Brugada syndrome. The honest statement is "no pathogenic variant detected across genes X, Y, Z; residual risk ~N%" — and for FH, ~10% of causal LDLR mutations are large deletions requiring MLPA, which no standard method here provides.

Call	Her genotype	Why it holds
Lactose intolerant	MCM6 rs4988235 GG · rs182549 CC · rs41380347 AA	Non-persistent at the European, Arabic/Bedouin and African persistence sites. Checking the Arabic allele was essential for a Levantine subject and it confirms rather than rescues the call. Best call in the report.
Earwax: wet	ABCC11 rs17822931 CC	Near-deterministic, >95% concordance
Power capacity: low	ACTN3 577XX (via proxies r²>0.98)	The best-replicated finding in sports genetics
Sleep depth: deep sleeper	ADA rs73598374 CT	ADA*2 allele, 6.2% of Europeans. Bachmann 2012 is literally titled "Functional ADA Polymorphism Increases Sleep Depth."
Caffeine sensitivity: higher	ADORA2A rs5751876 TT · CYP1A2 rs762551 AC	TT is the replicated caffeine–anxiety/sleep-disruption genotype
Vitamin A: higher need	BCO1 rs7501331 TT	Homozygous for reduced β-carotene→retinol conversion
NAFLD: elevated	PNPLA3 rs738409 CG · GCKR TC · TM6SF2 CC	Heterozygous at the strongest common NAFLD variant. The only well-founded disease call in the report.
Pain sensitivity: high	COMT APS/HPS · SCN9A AG · OPRM1 AG · FAAH CC	Four independent loci pointing the same way. Weak effects, genuinely concordant.
Alzheimer's: average	APOE ε3/ε3 (six concordant proxies + rs7412 CC)	Neither ε4 risk nor ε2 protection. Reassuring, relevant to the Dementia tab — and never named.
Carb / fat sensitivity, omega-3, appetite, inflammation	FTO all non-risk · TCF7L2 CC · FADS het · CRP het	All correct — and several understate how favourable her genotype actually is

Act V

The highest-confidence result in her genome is never named.

NAT2 slow acetylator. Homozygous, unambiguous, and absent from 85 pages.

Chapter 05 · Drug response

Twelve wrong dosing calls from one wrong gene assignment.

All six proton-pump inhibitors were marked "increase starting dose." She carries no CYP2C19*17 allele. And the report contradicts itself: it calls clopidogrel and escitalopram "use as directed" off the same gene.

NAT2*6A/*6A

Slow acetylator, homozygous at both defining SNPs so phase is unambiguous. The highest-confidence pharmacogenomic result in her entire genome — and it appears nowhere.

Isoniazid · sulfamethoxazole · procainamide · dapsone · sulfasalazine · hydralazine

Take this to a doctor before any surgery or TB screening

NAT2 *6A/*6A — slow acetylator. rs1041983 TT + rs1799930 AA + tag rs1495741 AA, all homozygous. It governs isoniazid (hepatotoxicity and neuropathy — and she is of Middle Eastern origin, where TB prophylaxis is a foreseeable exposure), sulfamethoxazole/co-trimoxazole, procainamide, dapsone, amifampridine and sulfasalazine.

CircleDNA flagged hydralazine "use with caution" — which is the right answer for this reason — but attributed it to nothing, so the finding is unusable. It then rated her "Detox: toxin generation speed → Normal," which is the opposite of slow acetylator.

The dosing calls, adjudicated

Drug / call	The report says	Her genotype	Correct position
Omeprazole, esomeprazole, lansoprazole, dexlansoprazole, pantoprazole, rabeprazole	Increase starting dose	rs12248560 CC — no 17. Full star-allele workup across all eight Tier-1 alleles gives 1/*1, normal metaboliser	CPIC: normal metaboliser → standard dose. Six drugs, mis-called twice each = 12 wrong calls
Morphine	Decrease dose	OPRM1 rs1799971 AG	The 118G allele associates with higher opioid requirement. Direction reversed — and CPIC reviewed OPRM1 and declined to issue a recommendation
Propofol	Decrease dose	No MH variant found	No genotypic basis, and mechanistically backwards — propofol is considered MH-safe. A hesitant anaesthetist is the actual risk created here.
Metformin	Use with caution ×3	Small-effect GWAS response hits	No CPIC, DPWG or FDA pharmacogenomic guidance exists for metformin. First-line for T2DM and PCOS in a woman of 40 — an unjustified flag risks steering her off an appropriate drug
Warfarin	Use as directed	VKORC1 rs9923231 CT (−1639 G/A het) + *CYP4F2 rs2108622 TT (3/3)* · CYP2C9 1/1	Effects partly offset, so net dose may land near standard — but this is CPIC Level A and FDA-labelled. Printing "as directed" and suppressing the diplotype destroys the actionable information
Statins (simvastatin)	Use as directed	SLCO1B1 rs4149056 TT = 1A/1A	Correct, and genuinely reassuring for myopathy risk

The workbook's Drug Response tab states the categories couldn't be extracted. They are printed in plain text on pages 55–61 under USE AS DIRECTED / DECREASE STARTING DOSAGE / USE WITH CAUTION / INCREASE STARTING DOSAGE. Seventeen drugs are not "use as directed" — including buprenorphine, fentanyl, methadone, morphine and propofol. The opioid and propofol entries are the kind of thing an anaesthetist needs before surgery.

9p21 · rs10757278 GG

Homozygous for the coronary risk haplotype — rated "average."

Corroborated by rs1333049 CC, rs1333048 CC, rs1537378 GG. The most replicated cardiovascular locus in human genetics, ~1.6–1.9× for early-onset coronary disease in homozygotes.

Independent of cholesterol, blood pressure and diabetes — which is exactly why her excellent lipids do not cancel it. Not alarm: a reason to keep her numbers where they are, and to mention it alongside family history.

Factor V Leiden · rs6025

Never tested. DVT rated "average."

Genuinely absent from the array — 60 other F5 SNPs are present, the nearest typed marker sits 63 bp away and is a different variant, and candidate proxies are r²≈0.2–0.3.

Carried by ~5% of Europeans and specifically actionable for a woman — combined oral contraceptives, HRT, pregnancy, surgery. Prothrombin G20210A is a genuine negative (rs1799963 GG). FVL status is undetermined, not reassured.

CYP2D6 · structural limit

Copy number cannot be read by any SNP array.

*5 deletions, *1xN/*2xN duplications and hybrid alleles are invisible. CYP2D6 governs ~20 drugs on CircleDNA's psychiatric and pain panels — meaning every CYP2D6-based statement in the report rests on data that cannot establish it.

The reassuring negatives she was never given credit for.

Several of these are higher-stakes than anything the report chose to headline.

3 LAYERS

DPYD activity score 2.0: Normal 5-FU / capecitabine dosing — among the highest-stakes pharmacogenomic results that exist
G6PD normal: Including the Mediterranean allele, the dominant deficiency variant in her ancestry
SLCO1B1 *1A/*1A: Normal function — statin myopathy risk not elevated
TPMT / NUDT15: Both normal
HLA tags: No HLA-B*57:01 / B*58:01 / A*31:01 · no HLA-C*06:02
Thrombophilia: Prothrombin G20210A negative (rs1799963 GG)
HFE: All wild-type — no haemochromatosis
LRRK2 G2019S: Negative — a genuinely valuable result given her ancestry
SERPINA1: MM — normal alpha-1 antitrypsin
Missed but useful: UGT1A1 *1/*28 — Gilbert carrier. Explains benign unconjugated hyperbilirubinaemia and spares unnecessary hepatology workups. Also ABCG2 Q141K het.

Absent from the array and material to a call

rs6025 (Factor V Leiden) · rs429358 (APOE ε4 — resolved by six proxies) · rs1815739 (ACTN3 — resolved by two proxies) · CFTR F508del (carrier status not excluded) · GBA1 N370S/L444P · MC1R R151C · FLG variants · MMP1 indel · SLC6A4 5-HTTLPR · DRD4 VNTR · AR CAG repeat. VKORC1 rs9923231 was present only under a suffixed probe name.

Structurally untypeable by SNP array regardless of coverage

CYP2D6 copy number · UGT1A1 *28 TA-repeat · HLA typing · RYR1 malignant-hyperthermia variants — MH is not excluded.

1 · NAT2 slow acetylator (*6A/*6A): Highest-confidence, most actionable result in her genome. Relevant to isoniazid, sulfamethoxazole, hydralazine, procainamide, dapsone. Never named in 85 pages.
2 · 9p21 homozygous coronary risk: Most replicated CV locus in genetics, rated "Average," structurally invisible to whole-exome sequencing.
3 · Factor V Leiden untested: Genuinely actionable for a woman around contraception, HRT, pregnancy and surgery. Undetermined — not negative.
Caveat that applies to all three: Any pharmacogenomic result would need confirmation in a CLIA/CAP laboratory before being used clinically. These are lines to open a conversation with, not instructions.

Act VI

Forty-three pages nobody opened.

And four of them can't be read at all without the colour original.

Chapter 06 · Microbiome

Eleven keystone species low. Butyrate normal.

The picture is genuinely mixed, and that matters — because the summary she was given presented only the deficits, which makes a partial problem look like a global catastrophe.

36%

Prevotella copri occupies 36% of her entire microbial community. Eleven keystone Lactobacillus and Bifidobacterium species are low or at 0.0%.

Acetate low · propionate low · butyrate NORMAL · 8 vitamins flagged malabsorbed

Keystone species low

Several at 0.0%

Butyrate

Normal

The one that matters most

Akkermansia

4.6×above range

Roseburia 1.7× above

Unreadable pages

Colour-coded only

What is genuinely low

The probiotic genera, and two of three short-chain fatty acids.

Eleven keystone Lactobacillus/Bifidobacterium species low or at zero — including L. salivarius and L. amylovorus at 0.0%. Acetate and propionate both low. Eight vitamins flagged as malabsorbed.

Gut Microbiome report · 43 pp

What is normal or better

The butyrate producers are intact.

F. prausnitzii, Roseburia, R. bromii and Akkermansia are all normal or above — Akkermansia 4.6× above the upper limit. Butyrate itself is normal. B2, B7, acetylcholine and histamine are unaffected.

Same report, same pages

Both halves are true. A summary that lists only the first half is not a summary — it is a sales page. See Act VII.

Four pages cannot be read, in either direction

Pages 18, 19, 22 and 23 encode pathogen levels, antibiotic resistance, post-antibiotic resilience and disease-risk flags in colour only. Text extraction cannot recover them.

So any claim about "high pathogen load" or "poor antibiotic resilience" is currently unverifiable either way — including the claims made in the clinic summary. Get the colour PDF. It is the cheapest open item in the whole folder.

Act VII

It cites a genotype she doesn't have.

Right conclusion, wrong allele — which means the reasoning can't be trusted anywhere.

Chapter 07 · The clinic doctor's summary

Dismissed — and here is why, item by item.

Four pages. It contradicts her genotype, contradicts the gut report it claims to summarise, and contradicts itself on the same page twice.

TT→ actually CC

Page 1 states she carries the "farmer" variant rs7903146-TT. Her raw genotype is CC — the favourable homozygote. TT is the risk genotype.

Wrong allele · contradicts the gut report · contradicts her genetics · contradicts itself

Contradicts the gut report it claims to summarise

The sheet says	The report actually says
"Gut–skin axis unaffected"	p10: reduced levels associated with skin health. A flat reversal — and he read the identical boilerplate sentence on p11 as a positive hair-health risk. Her 0.0% L. amylovorus is also linked on p16 to psoriasis and eczema.
"Vitamin malabsorption: A, B1, B3, B5, B6, B9, B12, C, D, K" — ten	The report flags eight. Vitamin A and B6 were assessed and NOT flagged — then both appear in his "prioritize" supplement list.
His supplement list	Vitamin B3 malabsorption is real and was dropped from it.
"High pathogen load: E. coli, C. difficile, Shigella dysenteriae, Candida tropicalis"	The report gives no level for any pathogen; none appear in any abundance table. Only three species get recommendations, and Shigella is not one of them.
"Mold exposure (linked to Aspergillus species detected)"	The words mold / mould / mycotoxin appear nowhere in 43 pages. The only Aspergillus with a value is A. chevalieri at 0.003% — the detection floor, same as ten other trace fungi, a normal food-storage fungus. Total fungal fraction: 0.086%.
"Low keystone probiotics" — lists six	The report lists eleven. Omits L. salivarius (0.0%), L. mudanjiangensis, L. fermentum, L. mucosae, L. ruminis.
Never mentions	That butyrate is normal, that F. prausnitzii, Roseburia, R. bromii and Akkermansia are all normal-or-above, or that B2, B7, acetylcholine and histamine are unaffected.

Presenting only the deficits makes the picture look globally catastrophic when it isn't.

Against three independent findings

"High protein: eggs, lean meats, legumes, protein shakes."

TFAP2B rs987237 AG — she regained 1.84 kg more per G allele on a high-protein diet. The Diet & Nutrition panel: "worse protein metabolism… may thrive on a low-protein diet." TCF7L2 CC: thrives on carbs.

It also contradicts his own page 1, which sets protein at 15%.

Against confirmed intolerance

Recommends kefir, yogurt and whey protein.

Against lactose intolerance confirmed four times over, higher casein sensitivity, and milk allergy at the 83rd percentile — and against the gut report's own p19 instruction to avoid dairy products.

Against itself

Barley appears on both lists.

Page 1 lists barley as a gluten grain to avoid. Page 3 recommends barley as a whole grain.

Macro split is 50/35/15 on page 1 against the source panel's 50/30/20. Neither is cited.

Unfounded additions

Recommendations with no source at all.

A "Celiac–gluten sensitivity" section (her tTG is negative and she carries no DQ2.5). L. amylovorus — not on the report's own strain list and not commercially available. L. helveticus — which the report calls normal. "12 weeks minimum" and PPI avoidance appear nowhere in the report.

Framing

Every page is headed "based on your DNA test result."

Including pages 3 and 4, which are entirely stool-microbiome content and have nothing to do with her DNA.

Why the genotype error matters

Right answer, wrong reason, unusable.

Some of the advice happens to land correctly. But when the stated evidence is a genotype she doesn't carry, there is no way to tell which parts were reasoned and which were guessed.

Act VIII

What survives all of it.

Four findings confirmed by two or more independent sources, and a Mediterranean diet named outright.

Chapter 08 · The parts that hold

You thrive on carbs. Nobody told you.

TCF7L2 rs7903146 CC is the favourable homozygote at the best-validated carbohydrate-response locus in the genome — and her workbook rated carbohydrate sensitivity "normal," understating it in her favour.

4× confirmed

Lactose intolerance and caffeine sensitivity are each confirmed four times over — CircleDNA, Weight, Diet & Nutrition, Food Sensitivities, plus the raw genotype. Those two are settled.

rs4988235 GG · rs762551 AC · rs7903146 CC · optimal diet named: Mediterranean

TCF7L2 rs7903146 CC · carbohydrates

Better carb response than average.

The panel is explicit: better response to carbs, may thrive on a high-carb diet, ~50% of calories from carbohydrate.

TFAP2B + panel · protein

Worse protein metabolism, not better.

She regained 1.84 kg more per G allele on high-protein. The panel independently says "may thrive on a low-protein diet." Two sources, one direction.

APOA2 + 42 variants · fat

Saturated fat specifically worse. MUFA typical.

The caveat the workbook inferred from APOA2 is now independently confirmed. And the panel names her optimal diet outright: Mediterranean.

MTNR1B rs10830963 CG + CRY2 rs11605924 AA

Late dinners hit her glucose harder.

Two circadian blood-sugar variants, both confirmed in the raw file. Implies time-restricted eating helps and the last meal should move earlier.

A free intervention that was never surfaced.

The dairy triad

Three separate adverse dairy axes, not one.

Lactose intolerance (confirmed ×4) · casein sensitivity higher · milk allergy more likely, 83rd percentile of 801 variants. The workbook carries only lactose.

This collides directly with the panel's "increased calcium need."

Sweet tooth · conflicted

More likely to prefer sweets — but not to binge.

TAS1R2 rs3935570 GG + FGF21 rs838133 AG across 40 variants: sweet preference and snacking both more likely. Offsetting: overeating, binge eating and eating disorders all less likely.

The nutrition tab, re-adjudicated · only one "higher need" survives

Nutrient	CircleDNA	Diet & Nutrition panel	Position
Vitamin D	Higher	Typical	Measured 71.37 — sufficient. Moot.
Vitamin A	Higher	Typical	CircleDNA right — BCO1 TT
CoQ10	Higher	Typical	NQO1 GG — full activity. Drop it.
Selenium	Higher	Typical	GPX1 GG — high activity. Drop it.
Vitamin E	Higher	Typical	CYP4F2 TT → higher circulating E. Drop it.
Iodine	Higher	Increased	Only agreement — but rests on a single SNP (GPX1), a selenium-pathway gene, while the same panel calls selenium typical
Iron	Normal	Increased ≥18 mg	Ferritin 10.8 — the panel is right
Calcium	Normal	Increased ≥1000 mg	Collides with the dairy triad — and with iron timing
Niacin / B3	Normal	Higher	Rests on 2 SNPs in SEPTIN8 and SLITRK1 — neither has a role in niacin metabolism. Keep "Normal."
Antioxidants	Normal	Splits	Glutathione increased function · lutein low · lycopene low
12 of 24 nutrients agree outright. 4 were never tested by CircleDNA at all. Of six "higher needs," only vitamin A holds — and it was already sufficient on the one that mattered.

MorningIron windowNothing competing

If iron is prescribed: alternate-day dosing, away from calcium, away from coffee and tea. Vitamin C alongside — not banned. Her TMPRSS6 genotype is precisely why the alternate-day evidence is relevant to her.

With mealsNo coffee or teaPolyphenols block iron

The "1–3 cups black coffee daily" recommendation stays — it just moves away from food. Caffeine sensitivity is confirmed ×4, so an earlier cut-off helps sleep as well as iron.

MealsMediterranean, carb-forwardNamed by the panel

~50% carbohydrate, MUFA-led fat, moderate protein. Saturated fat is the specific problem, not fat generally. Olive oil, fish, legumes, whole grains.

DairyThree reasons, not oneLactose · casein · milk protein

Calcium still needs to come from somewhere — greens, tahini, sardines, fortified alternatives — and needs to be timed away from iron.

EveningEarlier dinnerMTNR1B + CRY2

Two circadian glucose variants mean late carbohydrate is handled worse. Moving the last meal earlier is free, evidence-backed and specific to her genotype.

OngoingDo not carry forwardThree labels to retire

"Worse HLA-DQ genetics" (serology negative, rs2187668 CC) · the chilli/spice corroboration (2 of 3 risk genotypes absent) · "higher niacinamide needs" (2 irrelevant genes). Verify before use: the MTHFR rs1801131 GG reading — GG is the 1298C homozygote, conventionally read as reduced activity, but the panel calls it typical.

Act IX

Everything else, on the record.

The counting errors, the genotypes worth keeping, and the eight-item fix list.

Chapter 09 · Corrections and counts

Small errors, but they're the ones that compound.

Truncated variant lists, inverted numbers, and a "complete" summary that omits several non-baseline results.

Eight things to fix, in priority order — starting with the bloods and the ferritin, ending with a phone call to get one PDF re-sent in colour.

35 cancers · 103 conditions / 150 genes / 161 rows · 39 ACL SNPs · 19 ancestry rows

Counting and completeness errors inside the workbook

Where	It says	It should say
Cancer risk	34 cancer types	35 — the PDF and its own Cancer tab both say 35. Internally inconsistent.
Family planning	"All 150 conditions tested"	150 is the gene count. The PDF has 103 distinct conditions across 161 condition→gene rows. The tab footer inverts the same two numbers.
Ancestry	9 rows	10 of the PDF's 19 rows were dropped — all the 0% East/Southeast Asian lines. Values that are carried match to the decimal.
Fitness cross-check, ACL	"assessed 21 SNPs"	The PDF's ACL table runs to 39 variants across pp. 50–51. 21 is the first page only.
Weight tab, PLIN	"7 variants" — lists 4	Missing rs6496589 CC, rs8887 CT, rs2304796 GA — all three discussed on the source page, two of them favourable.
Weight tab, AMY1A	"9 variants" — lists 4	Same truncation pattern.
Red Flags summary	"every non-baseline result"	Omits Detox: cruciferous increased, Taste: super-taster, and 7 physical traits (facial/body hair, hair thinning, body odour, pain sensitivity, ear protrusion, smell sensitivity, thrill-seeking).
Sports & Fitness header	"Optimal sports: rowing, rock climbing"	The PDF reads "Rowing, rock climbing," with a trailing comma — the list is truncated, not closed. And "Low Power / High Endurance / High Strength" sits under Optimal Sports Type, not Optimal Training Type.
Data-quality warning	Partial mislabel list	Four more gene-column mislabels in the Weight PDF (PER2→HES6/ASB1, UCP3→UCP2, PLIN rs8887→UBXN6, ADIPOQ rs182052→RFC4), four in Diet & Nutrition (APOA2→FCER1G, CYP2R1→COPB1, DHFR→MSH3, CYP1A2→LMAN1L), plus the same rsID (rs3135350) assigned to two different genes across two reports issued the same day. The "key on rsID, not gene name" rule is right; the enumeration behind it is partial.
Fitness tab, HRR	"Slower"	The PDF says "predisposed to lower HRR." Same direction, paraphrased — the only one of 38 that isn't a literal match.

The fix list, in priority order.

Eight items. The first two are the only ones that touch her health.

8 ITEMS

1 · Add the bloods: They close the #1 priority (metabolic syndrome) and overturn the vitamin D flag. Belly fat still needs a tape measure.
2 · Ferritin 10.80: Flag iron deficiency. The one genuinely new actionable abnormality in the folder.
3 · Fix the Drug Response tab: From pages 55–61. Opioids, propofol and the PPIs are not "use as directed."
4 · Add the panels that exist: Diet & Nutrition (267 pp — it is not missing), and give Food Sensitivities its own tab.
5 · Correct the trait calls: Super-taster, sweet tooth, carbohydrate sensitivity. Add oxalate and the dairy triad to Red Flags.
6 · Fix the counts: 35 cancers · 103 conditions / 150 genes / 161 rows · 39 ACL SNPs · 19 ancestry rows · PLIN 7 and AMY1A 9 variant lists.
7 · Do not carry forward: "Worse HLA-DQ genetics" · the chilli/spice corroboration · "higher niacinamide needs." Verify before use: MTHFR rs1801131 GG.
8 · Get the colour gut PDF: Pages 18, 19, 22, 23 are unreadable without it — and they are the pages the clinic sheet makes its strongest claims about.

Gene / locus	Genotype	Why it's on this list
NAT2 6A/6A	rs1041983 TT · rs1799930 AA	Slow acetylator — isoniazid, sulfamethoxazole, hydralazine, procainamide, dapsone
9p21	rs10757278 GG	Homozygous coronary risk haplotype
F5 rs6025	not typed	Factor V Leiden undetermined — matters for contraception, HRT, pregnancy, surgery
TMPRSS6	rs855791 AG (+2 more het)	Blunted iron absorption — relevant to dosing schedule
HFE	all wild-type	No haemochromatosis — repletion is safe
VKORC1 / CYP4F2	rs9923231 CT · rs2108622 TT	Warfarin-relevant; CYP2C9 is 1/1
SLCO1B1	rs4149056 TT	1A/1A — normal statin transport
DPYD	activity score 2.0	Normal 5-FU / capecitabine metabolism
G6PD	normal, incl. Mediterranean allele	Relevant to her ancestry
APOE	ε3/ε3	Neither risk nor protection
MCM6	rs4988235 GG · rs41380347 AA	Lactose non-persistent at European and Arabic sites
TCF7L2	rs7903146 CC	Favourable — good carbohydrate response
PNPLA3	rs738409 CG	NAFLD-relevant; LFTs currently normal
MTNR1B · CRY2	rs10830963 CG · rs11605924 AA	Late meals handled worse

Germline genotype does not change. This table does not need re-testing — it needs to be in her medical record.

Open fronts

One blood draw and one appointment close almost all of it.

Front 1

Ferritin

Repeat with iron, TIBC, saturation, CBC and CRP — in one draw. Then find the cause.

Front 2

The record

NAT2 slow acetylator · 9p21 homozygous · Factor V Leiden untested. Three lines, written down once.

Front 3

A tape measure

The only metabolic criterion never measured — and the last thing keeping "belly fat" unresolved.

Across four DNA reports, twelve files and roughly five hundred genetic claims, exactly one new, measurable, correctable abnormality turned up. It was sitting in the folder unread while the workbook prioritised a metabolic syndrome score her bloods had already disproven.

Her genetics explain why she is vulnerable to it and confirm it is safe to treat. Coeliac, the cause most worth excluding, is already excluded. What's missing is the cause, and a current set of numbers.

The genome was never the problem. The reading of it was.

Neotrium Bio-Architecture · 679,568 SNPs · Not a diagnostic instrument · Not medical advice — genetic risk is not diagnosis, blood results carry dates, and the ferritin is from 27 April 2026. Discuss with a doctor.