APOE in Your 23andMe or AncestryDNA Raw Data: How to Read rs429358 and rs7412

Your 23andMe or AncestryDNA raw file contains the two variants that define APOE status, rs429358 and rs7412, whether or not you paid for a health report. Search the file for both rsIDs, read the pair off the table below, and you have your ε2/ε3/ε4 genotype. Before you do, read the next paragraph: this is a result many people wish they had not looked up.

Before you look. This result changes how people feel about their future, cannot be unlearned, and says something about your parents, siblings and children, who share your alleles. The Alzheimer's Association's position is that anyone considering genetic testing "should receive genetic counseling before a test is ordered" (alz.org, checked 14 October 2026). In the REVEAL trial, 162 adult children of Alzheimer's patients were randomized and 111 of them learned their APOE result with counselling: on average no significant short-term anxiety or depression followed, but those already distressed beforehand had the hardest time (Green et al., NEJM 2009). If that could be you, talk to a genetic counsellor first. The file will still be there next month.

What APOE ε2, ε3 and ε4 are#

APOE is a gene on chromosome 19 that codes for apolipoprotein E, a protein that carries cholesterol in the blood and the brain. It comes in three common alleles: ε2, ε3 and ε4. Worldwide, roughly 8% of alleles are ε2, 78% ε3 and 14% ε4 (Liu et al., Nature Reviews Neurology 2013). You carry two, one from each parent, so your genotype is a pair such as ε3/ε3, ε3/ε4 or ε4/ε4.

ε4 is the strongest common genetic risk factor for late-onset Alzheimer's disease; studies also report associations with higher LDL cholesterol and coronary artery disease, which is why it appears in several categories of a GWAS-based report. ε2 is associated with lower Alzheimer's risk and lower LDL. None of this says whether you will get a disease; the odds-ratio section below says how much it shifts the odds.

The two SNPs: rs429358 and rs7412#

The three alleles differ at two positions, 138 bases apart:

• rs429358: chromosome 19, position 45,411,941 on build GRCh37, the build 23andMe and AncestryDNA files use. The reference base is T; C defines ε4 (Ensembl GRCh37, SNPedia rs429358, checked 14 October 2026).
• rs7412: chromosome 19, position 45,412,079 on GRCh37. The reference base is C; T defines ε2 (Ensembl GRCh37, SNPedia rs7412).

The table uses the letters exactly as a 23andMe file prints them: 23andMe reports every genotype on the plus strand of GRCh37 (its file header says so, and SNPedia's orientation page says the same), both SNPs are plus-strand variants in dbSNP, so nothing needs flipping, and the two letters of a call are printed in alphabetical order, so expect CT rather than TC.

The mapping is SNPedia's APOE genoset table; the ε3/ε3 frequency is from its gs246 page, and the ε4 figures follow the Alzheimer's Association's US estimates that 20–30% of people carry one or two copies and about 2% carry two (alz.org, checked 14 October 2026). The CT/CT row is a limitation of array data: a chip cannot tell whether the two heterozygous calls sit on the same chromosome or on opposite ones.

How to find them in your file without uploading it#

Three steps, no website needed:

1. Open the file in a plain text editor. Unzip the download (see our download guide) and open the .txt (23andMe, AncestryDNA) or .csv (MyHeritage) in TextEdit, Notepad or a code editor, not Excel, which can reformat values.
2. Search for rs429358. A 23andMe line looks like rs429358 19 45411941 CT, with the genotype as one two-letter field; an AncestryDNA line splits the alleles into two columns, C and T. Note the letters, then search for rs7412.
3. Read the pair off the table above. If a line shows -- (23andMe) or 0 0 (Ancestry), the array did not call that position and the file cannot answer.

Our guide to reading a raw DNA data file explains the columns, builds and no-calls. SNPedia's chip tags list both SNPs on 23andMe's v3 to v5 chips and Ancestry's v2 chips, so a file from either company should contain them.

Why an AncestryDNA file may show rs429358 as TT for everyone#

There is a long-standing community report that AncestryDNA files miscall rs429358. SNPedia's genoset page for ε3/ε3 states that "Ancestry data always reports rs429358 as (T;T), even for people who's data from other sources indicates they are (C;T)", calls it a false-negative problem, and warns that APOE genosets assigned from Ancestry data, including by Promethease, will be wrong for some users who are really ε3/ε4 or ε4/ε4 (SNPedia gs246, last edited 4 January 2018, checked 14 October 2026). We found no statement from Ancestry confirming or fixing it and have not verified it on a 2026 file, so treat it as a reported issue. If it holds, the error is one-directional: an Ancestry TT can hide ε4; a CT or CC is not a false positive in the same way.

A second caveat is newer. Genetic Lifehacks reports that AncestryDNA raw files changed in early March 2026 and now contain far fewer variants (Genetic Lifehacks, single source, checked 14 October 2026). We do not know whether rs429358 and rs7412 survived the change; if your file is from after March 2026 and the lines are missing, that is why.

If an APOE result matters to a decision, a consumer array file is the wrong place to settle it; a clinical laboratory test, ordered and read with a genetic counsellor, is.

What the odds ratios mean in absolute terms#

A GWAS row reports an odds ratio, and an odds ratio is not a probability. For APOE the numbers are large enough to be worth translating. Relative to ε3/ε3, a large 1997 meta-analysis of people of European ancestry found an odds ratio of about 3 for one copy of ε4 (ε3/ε4: 3.2) and 12 to 15 for two (ε4/ε4: 14.9), with smaller effects in people of African and Hispanic ancestry (Farrer et al. 1997, summarized in Liu et al. 2013). A later analysis of about 7,500 cases and 10,132 controls found 3.6 and 14.5, and 0.56 for ε2 carriers (Genin et al., Molecular Psychiatry 2011); 23andMe's own article on its Alzheimer's report uses the bands "around threefold" and "eight- to 12-fold" (23andMe, 2 December 2022, checked 14 October 2026).

Genin and colleagues also converted the odds into lifetime risk by age 85:

One copy of ε4 roughly triples a risk that starts under 10%, so about three in four carriers reach 85 without the disease. Two copies is a different situation: around half are affected by 85, the average age of onset shifts from 84 in non-carriers to 68 (Liu et al. 2013), and a 2024 study of 3,297 autopsied brains and 10,039 living participants found Alzheimer's biology in nearly all ε4/ε4 carriers by their mid-sixties, leading its authors to propose ε4 homozygosity as a genetic form of the disease (Fortea et al., Nature Medicine 2024). Between 40% and 65% of people diagnosed with Alzheimer's carry ε4 (alz.org), which also means a third or more do not. These figures come from cohorts of mostly European ancestry, and they describe groups, not you.

How the Full Human neurological and mental health report shows it#

Our neurological and mental health report (the sample report shows the same tables) does not print "ε3/ε4". It prints the GWAS Catalog rows your file matches. The health catalog we ship (release 2026-06-01) has 482 rows mapped to Alzheimer's disease across many variants; rs429358 appears in 71 rows across our health and nutrition catalogs, 57 of them in the health catalog, under Alzheimer's disease, dementia with Lewy bodies, type 2 diabetes and myocardial infarction, among others, and rs7412 in 15 rows, including Alzheimer's disease and coronary artery disease.

Each row shows the rsID, the risk allele the study reported, your genotype, the odds ratio and the PubMed ID, as explained in how to read the health report. A study row counts only if your genotype contains the risk allele it reported, and the table groups matched rows into one line per rsID, genotype and risk allele, with the effect averaged per study and then across studies. Most Alzheimer's rows for rs429358 name C as the risk allele, with odds ratios between about 1.6 and 4.2; one 2025 study names T (odds ratio 3.1) and another reports five protective T rows (0.42 to 0.61). So a CT or CC file matches the C rows, a TT file matches only the T rows, and even an ε3/ε3 report shows an rs429358 line with risk allele T under Alzheimer's; a CT file shows two lines, one per allele. The printed effect is the average of the studies behind that line, not the range. Rows whose study reported no risk allele ("?") never match and are not scored.

The per-condition verdict ("higher", "lower" or "average or undetermined") multiplies the odds ratios of every matched row for that condition. As the methodology page spells out, it does not weigh one copy against two, does not correct for correlated variants and uses no baseline risk: it is an association, not a polygenic risk score and not a diagnosis. For an ε4 carrier the Alzheimer's verdict will usually read "higher"; the line that tells you something is the rs429358 row whose risk allele is C. Read the rows, then take the page to a doctor or genetic counsellor. Nothing leaves your device: the file is parsed in your browser tab and the PDF is written there, which you can check yourself.

FAQ#

Does 23andMe test for APOE4?

Yes, in two ways. 23andMe's Late-Onset Alzheimer's Disease Genetic Health Risk report, one of ten reports the FDA authorized on 6 April 2017 (23andMe press release, checked 14 October 2026), reports the ε4 variant and is sold only with the Health service in eight countries. Separately, the raw file every customer can download contains rs429358 and rs7412, so you can read the status yourself even with an ancestry-only kit.

What is the rsID for APOE?

There are two. rs429358 (chromosome 19, position 45,411,941 on GRCh37) defines ε4 when it reads C, and rs7412 (position 45,412,079) defines ε2 when it reads T. Neither alone gives your APOE genotype; you need both, read together from the table in this article. Searching a raw file for "APOE" finds nothing, because raw files list rsIDs, not gene names.

Can I have APOE4 and never get Alzheimer's?

Yes, and most people with one copy do not. In the lifetime-risk estimates from Genin et al. 2011, about 23% of men and 30% of women with ε3/ε4 develop Alzheimer's by 85, so roughly three in four do not. Two copies carry a much higher risk, around half by 85, but even there it is not a certainty. ε4 is a risk factor, not a cause on its own.

Should I find out my APOE status?

That is a personal decision, and for a healthy adult today no approved prevention depends on the result. The REVEAL trial found that learning the result with counselling did not cause lasting distress on average, but people who were already anxious struggled more. If you want to know, get genetic counselling first, decide in advance what you would do with each result, and remember that your relatives share the answer.