Learn Read your report
Confidence tiers on a DNA report: what Tier 1, 2 and 3 mean
A genome report can print a guideline-backed drug interaction and a contested personality gene in the same font. Ours sorts every health and trait section into three confidence tiers first. This is how the tiers are defined, which outside grades sit behind them, and what a tier never says.
Key takeaways
- The report has three tiers: Tier 1 · Established, Tier 2 · Well-supported and Tier 3 · Emerging / Speculative. Each is a fixed group of report sections, set in the report's code, and the tier does not change with what your file contains.
- Inside a tier, each card keeps its outside grade. In Tier 1 and the carrier sections that is ClinVar's star rating: zero-star classifications are never shown, one-star ones sit in a separate lower-confidence group, two or more stars make the main list[1].
- Tier 2 restates CPIC's prescribing guidance, which exists only for level A and B gene-drug pairs[2], alongside carrier status, traits that passed a replication gate, and polygenic disease scores.
- Tier 3 sits below a line marked credibility firewall. Its 92 Frontier and Fringe entries each carry a strength label, from established biology to popularly over-hyped[12].
- A tier grades evidence, never risk. A Tier 1 card is not a diagnosis, and a Tier 3 card is not a verdict that the finding is false.
Genetic findings rest on very different amounts of evidence, and the field already grades the difference. ClinVar gives each variant classification zero to four stars for how it was reviewed[1]. CPIC, the consortium that writes gene-based prescribing guidance, ranks gene-drug pairs from level A to level D, and only A and B pairs carry a recommended prescribing change[2]. Genome-wide studies call an association significant only at a p-value of 5×10⁻⁸ or lower[3]. The Aimosti report groups its sections into three tiers on top of these grades and leaves each grade on the card where it applies.
The three tiers, as the report prints them
A confidence tier is a heading that groups report sections by how well tested their science is. The report has three, and each heading carries a one-line banner. The table gives both exactly as they appear on a report, along with the sections each tier holds.
| Tier | Banner on the report | Sections it holds |
|---|---|---|
| Tier 1 · Established | Strong, replicated science, often recognised in clinical guidelines: the most trustworthy signals in your report. | Clinical findings, risk factors, APOE, Lp(a), APOL1, HFE, celiac HLA, the wider screen's dominant conditions |
| Tier 2 · Well-supported | Solid, reproducible findings: pharmacogenomic and carrier-grade. Reliable, though not all are clinically actionable. | Carrier status, the wider carrier screen, pharmacogenomics, Deep Read, traits, ABO blood type, polygenic disease scores |
| Tier 3 · Emerging / Speculative | Real but preliminary: mixed, contested, or population-limited. Interesting to explore, never to act on: the label is the point. | Frontier, emerging scores, the Fringe |
Source: The report's own tier labels and banners, as on the sample report[4].
The tier belongs to the section. The report's code assigns each section to one tier, and every card inside it stays there whatever the file contains; strength within a section is shown on the card. One consequence surprises people. A single-submitter BRCA1 match, the weakest clinical match the report shows, appears inside Tier 1, because it is a clinical finding, in a separate group marked lower-confidence below the main list.
A small chip repeating the tier label appears beside each group of the at-a-glance summary, on every polygenic score card and on the ABO blood type card; other cards take their tier from the heading above them. Ancestry, runs of homozygosity and archaic ancestry sit outside the tiers. A haplogroup or an ancestry share is a classification, and ranking it beside risk findings would blur what the tiers mean. Archaic ancestry is drawn in Tier 3's exploratory style all the same, because its claims about consequences are exploratory. The Neanderthal DNA guide explains what that section reads and why it gives no percentage.
Tier 1, and the ClinVar stars inside it
Most of Tier 1 is clinical findings: variants in the report's curated panel genes that ClinVar classifies as pathogenic or likely pathogenic, the two top classes of the five-term scale clinical laboratories use[5]. The report does not classify variants itself. What decides whether a match is shown, and where, is ClinVar's own review status.
| Stars | ClinVar's review status | In the report |
|---|---|---|
| 4 | Practice guideline | Main list |
| 3 | Reviewed by expert panel | Main list |
| 2 | Criteria provided, multiple submitters, no conflicts | Main list |
| 1 | Criteria provided, single submitter; or conflicting classifications | A separate lower-confidence group, labelled as leads |
| 0 | No assertion criteria provided | Never shown |
Source: Star definitions from ClinVar[1]. The same rule applies to clinical findings, carrier status and the wider carrier screen. A conflicting classification is not pathogenic or likely pathogenic, so it never reaches a clinical card.
The sample report shows the range in a single section[4]. Its familial hypercholesterolemia finding in LDLR has three stars, because an expert panel reviewed the classification. Its long QT syndrome finding in KCNQ1 has two: several laboratories submitted with evidence and agreed. Its BRCA1 match has one, a single laboratory's classification, and sits in the lower-confidence group under a note that the evidence is thinner.
The rest of Tier 1 is single-locus modules with their own evidence notes: APOE, lipoprotein(a), APOL1, HFE hemochromatosis and the celiac HLA rule-out. Then comes the dominant part of the wider screen, which reads exact ClinVar positions in 52 further genes. A gene gets there only if ClinGen, the Clinical Genome Resource, rates its link to the condition Definitive or Strong, and a variant only with two or more stars[6]. ClinGen's scale grades a gene's link to a disease, a different question from whether one variant is harmful. It runs from Definitive through Strong, Moderate and Limited to Disputed and Refuted[7, 8].
Tier 2: drug guidance, carrier status, traits and scores
Tier 2 holds sections whose evidence is solid but whose use is narrower. Pharmacogenomics restates CPIC's guidance. CPIC writes prescribing recommendations only for level A and B gene-drug pairs; for a level C or D pair, the evidence is not considered adequate for any recommendation[2]. The sample report's CYP2C19 card reads intermediate metabolizer for the *1/*2 diplotype[4]. CYP2C19 and clopidogrel is a level A pair[9], and CPIC's 2022 guideline reports that intermediate and poor metabolizers taking clopidogrel have reduced platelet inhibition and a higher risk of major cardiovascular and cerebrovascular events[10].
The level is printed where there is nothing to restate, too. The same card lists drugs CPIC reviewed with CYP2C19 without publishing a recommendation, each with its level; brivaracetam is level B, and the card says the gap is no sign the gene does not matter for it.
Carrier status uses the same star rule as Tier 1. The sample's cystic fibrosis carrier finding in CFTR has three stars[4]. The wider carrier screen covers recessive and X-linked conditions in 1,363 more genes, rated Definitive, Strong or Moderate by ClinGen (a short reviewed list aside), and admits variants with one star or more, the one-star ones again in a separate group[6]. One name clashes. The carrier panel is drawn from ACMG's carrier-screening list, whose Tier 3 means conditions with a carrier frequency of at least 1 in 200[11]. That ACMG tier is a list of conditions to screen for and has nothing to do with the report's Tier 3.
Traits pass a written gate. One drawn from genome-wide studies needs its lead variant at p ≤ 5×10⁻⁸, replication in two independent cohorts and a frequency of at least 1 percent; a functional trait such as earwax type needs a landmark discovery paper and a known mechanism instead. Of the 28 traits, 15 come through the genome-wide route, 12 through the functional one and one, caffeine metabolism, through pharmacogenetics[12]. The 17 polygenic disease scores sit here too, and each is placed on its distribution only when the file holds at least half of its scoring variants.
Below the firewall: Tier 3
The dashed line reads credibility firewall, exploratory below. Under it are the Frontier, with 60 entries, the Fringe, with 32, and nine emerging polygenic scores for traits that can be measured or asked about directly, such as height[12]. Each Frontier and Fringe card carries one of seven strength labels, and the chart counts them. One marker can appear on both sides of the line: the HERC2 eye-colour variant has a Traits card and a Fringe card (what it can and cannot say).
Two sample cards show what the labels are for. COMT rs4680 is labelled Mixed evidence[4]. The meta-analysis it cites pooled 12 studies of one card-sorting test, 1,910 people, and found a small advantage for Met/Met over Val/Val in healthy volunteers (d = 0.29), none in patients with schizophrenia, and larger effects in the earlier studies[13]. That decline is common: across 36 genetic associations, the first study often suggested a stronger effect than later ones[14].
The serotonin-transporter card, read at SLC6A4 rs25531, is labelled Contested and titled the famous null. A 2019 preregistered study tested 18 much-studied depression candidate genes in samples of 62,138 to 443,264 people and found no clear association and no interaction with life stress[15]. Its own summary says it is kept as the textbook example of a finding that did not replicate.
How the outside grades line up with the tiers
| System | What it grades | Its scale | Where the report uses it |
|---|---|---|---|
| ClinVar review status | How one variant's classification was reviewed | 0 to 4 stars | Decides which clinical and carrier matches are shown and in which group |
| ClinGen gene-disease validity | How firmly a gene is tied to a disease | Definitive, Strong, Moderate, Limited, Disputed, Refuted | Selects the genes of the wider screen |
| CPIC level | Whether genetics should change prescribing of a drug | A, B, C, D | Pharmacogene guidance comes from A and B pairs; C and D pairs are listed with their level |
| Genome-wide significance | Whether an association is likely real | p ≤ 5×10⁻⁸ | Part of the gate every genome-wide trait passes |
| Aimosti strength label | How settled a Frontier or Fringe reading is | Seven labels | Printed on every Tier 3 card |
Source: ClinVar[1], ClinGen[7, 8], CPIC[2], genome-wide significance[3, 16], and the report's own content[12].
Each system answers its own question, so the report keeps them apart. The genome-wide threshold has a simple origin. Testing a whole genome amounts to about a million independent tests in Europeans[16], and dividing the usual 5 percent error rate by a million gives 5×10⁻⁸, our arithmetic. Clearing it once is not the end, because first studies tend to overstate effects, as the COMT example showed[14]. The trait gate asks for replication as well.
None of these maps one to one onto a tier. A Frontier entry labelled established biology stays in Tier 3, and its label adds that the call is inferred through a nearby variant; a one-star clinical match stays in Tier 1, in its own group. The tier is the shelf. The grade on the card says how firm that one reading is.
What each kind of file can fill
The tiers are the same for every file; what fills them is not. Our guide to genome file formats explains how a chip export, a VCF, a gVCF and a BAM or CRAM differ.
| Section | Chip export | Plain VCF | gVCF | BAM or CRAM, added to a VCF or gVCF |
|---|---|---|---|---|
| Clinical findings (Tier 1) | 2 variants: Factor V Leiden and prothrombin G20210A | 39 conditions | 39 conditions, with the genes it read shown | The same |
| Carrier status and wider screen (Tiers 1 and 2) | Not available | 60 conditions and 1,415 more genes | The same | The same, plus 2 deletions called from read depth |
| Pharmacogenomics (Tier 2) | Up to 7 genes, as the chip allows | 9 genes | 25 genes | 29 genes, with CYP2D6 copy number |
| Polygenic disease scores (Tier 2) | 17 attempted; shown only with half the variants present | 17 | 17 | 17 |
| Frontier and Fringe (Tier 3) | Up to 85 entries, those the chip typed | 92 entries | 92 entries | 92 entries |
Source: Counted from the content the report loads, as listed on our coverage page[12].
A chip export loses the most from the top tiers: it cannot report carrier status, and its clinical section reads only the two clotting variants it can type by name. Its polygenic scores are attempted against the half-the-variants bar, and on the sample chip report none of the 17 cleared it[4]. A plain VCF fills every section but cannot prove that a missing variant was looked for. A gVCF can, and it adds 16 pharmacogenes. Aligned reads add copy number, deletion calls and a per-gene callability map. Coverage and callability explains how the report says which genes were read.
What a tier never means
A tier is not severity. Earwax type sits in Tier 2 alongside carrier findings, and the tier says nothing about which of them matters more to anyone's health. It is not a probability either. A polygenic score in Tier 2 is a ranking against a reference group, and what a percentile means covers why that is not a personal risk.
A tier does not vouch for the genotype call either. That depends on the file and is reported separately, as coverage: a Tier 1 section built from a plain VCF rests on strong science and still cannot show that an absent variant was checked. Tier 3 does not mean false; many of its entries are replicated and small. And a tier never moves. No section is promoted because a result looks striking, or demoted because it is unwelcome.
What Aimosti would (and wouldn't) show you
Every health and trait section sits under one of three tier headings with a one-line banner, and a dashed line marked credibility firewall separates Tier 3 from the rest. Each card keeps its own grade: ClinVar stars on clinical and carrier findings, CPIC guidance and levels on pharmacogenes, the trait gate on traits, a strength label on every Frontier and Fringe entry. Ancestry sections render after the tiers, outside them.
What we won't claim
We won't move a section into a higher tier to make the report look more certain, drop a tier heading or a strength label, or let a high tier stand in for a diagnosis. A tier grades the kind of evidence behind a section. It never says how likely anything is to happen to the person reading it.
Bottom line. The tier tells you how much the science behind a section has been tested, before you read the result. It is fixed by section in the report's code, it keeps the outside grade it rests on, and it says nothing about your own risk.
Questions people ask
Is Tier 1 more serious than Tier 3?
No. The tiers grade how well tested the science behind a section is. A Tier 1 risk factor such as Factor V Leiden can carry a low effect, and the report says so on the card[4].
Why is a one-star BRCA1 match in Tier 1 if the evidence is thin?
Because the tier belongs to the section, and clinical findings are a Tier 1 section. The star rating is the finer grade: one-star matches sit in their own lower-confidence group, described as leads to verify, below the two-star and higher list[1].
Does a Tier 3 label mean the finding is wrong?
No. Tier 3 is for readings that are real but preliminary, mixed, contested or limited to some populations. Of its 92 Frontier and Fringe entries, 41 are labelled replicated with a small effect, and a few are kept to say that a famous claim did not hold[12, 15].
Is the report's Tier 3 the same as ACMG Tier 3 carrier screening?
No. ACMG's tiers are lists of conditions for carrier screening, and its Tier 3 means conditions with a carrier frequency of at least 1 in 200[11]. The report's carrier panel is drawn from that list and sits in Tier 2 · Well-supported.
Can my file change which tier a section is in?
No. The tier is fixed per section. The file decides which sections can be filled: a chip export has no carrier section, for example, and a tier with nothing to show is left off the report.
References
- Review status in ClinVar. ClinVar documentation, NCBI. Read 2026-10-10. Aggregate records: four stars, practice guideline; three, reviewed by expert panel; two, criteria provided, multiple submitters, no conflicts; one, criteria provided, conflicting classifications, or criteria provided, single submitter; none, no assertion criteria provided.
- Assignment of CPIC levels for genes/drugs. Clinical Pharmacogenetics Implementation Consortium (CPIC), via ClinPGx. Read 2026-10-10 from the page's data endpoint: only level A and B pairs have sufficient evidence for at least one prescribing action to be recommended; level C and D pairs are not considered to have adequate evidence or actionability for prescribing recommendations.
- Fadista J, Manning AK, Florez JC, Groop L. The (in)famous GWAS P-value threshold revisited and updated for low-frequency variants. European Journal of Human Genetics, 2016. doi:10.1038/ejhg.2015.269
- Sample report: a made-up genome run through the real report engine. Aimosti, 2026. Read 2026-10-10, whole-genome view. Tier 1: LDLR familial hypercholesterolemia, three stars (reviewed by expert panel); KCNQ1 long QT syndrome, two stars; BRCA1, one star, in the lower-confidence group. Tier 2: CYP2C19 *1/*2 intermediate metabolizer; CFTR carrier, three stars. Tier 3: COMT rs4680 Met/Met, Mixed evidence; SLC6A4 rs25531, Contested.
- Richards S, Aziz N, Bale S, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genetics in Medicine, 2015. doi:10.1038/gim.2015.30
- Wider ClinVar screen: site table and build manifest. Aimosti, 2026. ClinVar 2026-07-20 release, selection rules of 2026-10-05. Dominant part: 52 genes, 7,956 sites (7,464 with two stars, 492 with three); 14,115 dominant-condition sites dropped for having fewer than two stars. Recessive part: 1,263 genes, 89,906 sites (62,953 one star, 26,146 two, 807 three). X-linked part: 100 genes, 8,416 sites (6,406 one star, 1,576 two, 434 three).
- Strande NT, Riggs ER, Buchanan AH, et al. Evaluating the clinical validity of gene-disease associations: an evidence-based framework developed by the Clinical Genome Resource. American Journal of Human Genetics, 2017. doi:10.1016/j.ajhg.2017.04.015
- ClinGen gene-disease validity curations (download). Clinical Genome Resource, 2026. File created 2026-10-10: 3,705 curations. Definitive 2,309; Strong 85; Moderate 460; Limited 555; Disputed 199; Refuted 49; No Known Disease Relationship 48.
- CPIC gene-drug pairs: CYP2C19 with clopidogrel. CPIC database. Read 2026-10-10: CYP2C19 and clopidogrel, CPIC level A, used for recommendation.
- Lee CR, Luzum JA, Sangkuhl K, et al. Clinical Pharmacogenetics Implementation Consortium guideline for CYP2C19 genotype and clopidogrel therapy: 2022 update. Clinical Pharmacology and Therapeutics, 2022. doi:10.1002/cpt.2526
- Gregg AR, Aarabi M, Klugman S, et al. Screening for autosomal recessive and X-linked conditions during pregnancy and preconception: a practice resource of the American College of Medical Genetics and Genomics (ACMG). Genetics in Medicine, 2021. doi:10.1038/s41436-021-01203-z
- Exactly what we analyse: every panel, gene and score in the report. Aimosti, 2026. Counts as of 2026-10-20: 39 clinical conditions (2 from a chip export), 60 carrier conditions, the wider screen in 1,415 genes, 9 pharmacogenes from a plain VCF (up to 7 from a chip export) and 25 from a gVCF, 29 from aligned reads, 28 traits, 17 polygenic disease scores, 9 emerging scores, 60 Frontier and 32 Fringe entries.
- Barnett JH, Jones PB, Robbins TW, Müller U. Effects of the catechol-O-methyltransferase Val158Met polymorphism on executive function: a meta-analysis of the Wisconsin Card Sort Test in schizophrenia and healthy controls. Molecular Psychiatry, 2007. doi:10.1038/sj.mp.4001973
- Ioannidis JP, Ntzani EE, Trikalinos TA, Contopoulos-Ioannidis DG. Replication validity of genetic association studies. Nature Genetics, 2001. doi:10.1038/ng749
- Border R, Johnson EC, Evans LM, et al. No support for historical candidate gene or candidate gene-by-interaction hypotheses for major depression across multiple large samples. American Journal of Psychiatry, 2019. doi:10.1176/appi.ajp.2018.18070881
- Pe'er I, Yelensky R, Altshuler D, Daly MJ. Estimation of the multiple testing burden for genomewide association studies of nearly all common variants. Genetic Epidemiology, 2008. doi:10.1002/gepi.20303
Last reviewed . Every number on this page links to the source it comes from; if one of them has moved, tell us.