Glossary

A result is written in the registries’ own words, because paraphrasing what a registry said is how a tool ends up reporting something nobody said. The cost is a page ofLOEUF, missense_variant andcriteria provided, multiple submitters, no conflicts. Here is what all of it means.

On a result, every one of these terms is underlined: hover it, tap it, or tab to it for the same definition in place. The switch above the result lists them under each section instead, and every export carries the definitions for the terms it uses.

These definitions explain what each term means in general, in plain English, and each links to the published definition it restates. None of them is a statement about the variant you looked up.

What a result is measured against

Where the change sits, and the map the position is counted on. None of this says whether a change matters; it says which change is being talked about.

Genome build: The reference map of the human genome that positions are counted against. GRCh38 is the current map and GRCh37 the one before it. The same spot in the genome has a different number on each map, so a position means nothing until you know which map it was measured on.Genome Reference Consortium
Position: Which chromosome the change sits on, and how far along it sits, counted in DNA letters from one end of that chromosome.Genome Reference Consortium
Gene: A stretch of DNA carrying the instructions for one product the body makes, usually a protein. Genes have short symbols, like BRCA1.NCBI Gene
dbSNP rsID: A permanent catalogue number for a place in the genome where people are known to differ. It always starts with “rs”. The number is a label and nothing more — it says only that the difference has been catalogued, not that it matters.dbSNP at NCBI
Reference sequence: The catalogued stretch of sequence a change is counted along, named in front of the description — NM_005957.5 in NM_005957.5:c.665C>T. A gene is often read out in more than one version, and the versions begin at different points, so one change lands on a different number in each. The number on its own does not say which count it belongs to.HGVS nomenclature: reference sequences
HGVS coding notation: The agreed way of writing where a change sits inside a gene’s coding instructions. It starts with “c.”, then the position, then what changed — c.5266dupC means a C was duplicated at position 5266. What comes before the colon is the sequence that position is counted along.HGVS nomenclature
HGVS protein notation: The same change written as its effect on the protein. It starts with “p.” and names the original building block, its position in the protein, and what took its place. What comes before the colon is the protein it is counted along.HGVS nomenclature
Filed as: The exact spelling a registry keeps a change under. Registries sometimes describe one change in different ways — the same duplication can be written against two neighbouring positions — so both spellings are shown rather than one being quietly chosen.HGVS nomenclature

What the change does to the protein

Predictions worked out from where the change falls in the sequence. They describe a mechanism, not an outcome in a person.

Consequence: A prediction of what the change does to the protein a gene codes for — swap one building block, cut the protein short, or change nothing. It is worked out from where the change sits in the sequence. Nobody was tested to produce it.Ensembl consequence terms
Missense variant: The change swaps one amino acid — one building block of the protein — for a different one. The protein is still made, at full length, with one block changed. Whether that matters depends on the block and where it sits.Ensembl consequence terms
Synonymous variant: A DNA letter changes but the protein comes out identical, because several DNA spellings stand for the same amino acid.Ensembl consequence terms
Frameshift variant: Letters are added or lost in a number that is not a multiple of three. The cell reads DNA three letters at a time, so everything after the change is read in the wrong groupings and the rest of the protein comes out garbled.Ensembl consequence terms
Stop gained: The change creates a stop signal where there was none, so the protein is cut short and the missing part is never made. Also called a nonsense variant.Ensembl consequence terms
Stop lost: The change removes the signal that says where the protein ends, so the cell keeps building past the proper finish.Ensembl consequence terms
Start lost: The change removes the signal that says where to begin building the protein.Ensembl consequence terms
Splice donor variant: The change sits at a join where the cell cuts an unused stretch out of the instructions and pastes the useful parts together. Changes right at a join can make the cell cut in the wrong place.Ensembl consequence terms
Splice acceptor variant: The change sits at the other end of one of those joins, where the cell picks the instructions back up after cutting a stretch out.Ensembl consequence terms
Splice region variant: The change is close to one of the joins where the cell cuts and pastes the instructions together, but not on the join itself.Ensembl consequence terms
In-frame deletion: Letters are lost in a multiple of three, so whole building blocks are removed and everything after them is still read in the right groupings.Ensembl consequence terms
In-frame insertion: Letters are added in a multiple of three, so whole building blocks are added and everything after them is still read in the right groupings.Ensembl consequence terms
Intron variant: The change is in a stretch that is cut out of the instructions before the protein is built, so it often leaves the protein itself untouched.Ensembl consequence terms
5′ UTR variant: The change is in the run-up to the coding instructions — copied by the cell, but not turned into protein. Changes here bear on how much protein is made rather than on its shape.Ensembl consequence terms
3′ UTR variant: The change is in the tail after the coding instructions — copied by the cell, but not turned into protein. Changes here bear on how long the message lasts and how much protein is made from it.Ensembl consequence terms
Upstream gene variant: The change sits outside the gene, before where it starts.Ensembl consequence terms
Downstream gene variant: The change sits outside the gene, after where it ends.Ensembl consequence terms
Non-coding transcript variant: The change is in a gene that is copied by the cell but never built into a protein. Some of those copies do a job of their own.Ensembl consequence terms
Coding sequence variant: The change is inside the part of the gene that codes for the protein, but the record does not say more precisely what it does there.Ensembl consequence terms

What submitters concluded, in ClinVar

ClinVar’s own vocabulary. Each label is one submitter’s conclusion about a change and a condition on a date, and the review status says how much checking sits behind it.

ClinVar: A free public archive run by the US National Institutes of Health, where labs and clinics deposit the conclusions they reached about genetic changes. ClinVar stores and shows what each submitter said; it does not test anything itself and does not arbitrate between submitters who disagree.ClinVar’s definitions
Submission: One submitter’s record of one change, assessed against one condition, on one date. Several labs can submit about the same change, and each keeps its own record. The identifier beginning “RCV” is that record’s permanent number.ClinVar identifiers
Variation ID: ClinVar’s own number for a change, shared by every submission about it. It is how ClinVar keeps one change together when submitters write it differently.ClinVar identifiers
Condition: The disease or trait the submitter judged the change against. One change can be judged against several conditions, and a submitter can reach a different conclusion for each.ClinVar’s definitions
Pathogenic: ClinVar’s label for a change a submitter concluded does cause the condition named next to it. It is that submitter’s conclusion about the change in general, recorded on a date — not a finding about any particular person, and not a statement that a carrier will fall ill.ClinVar’s definitions
Likely pathogenic: The submitter thinks the change causes the condition but the evidence stops short of certain. In the guidelines submitters follow, “likely” is meant to mean about 90% confidence.ClinVar’s definitions
Uncertain significance: There is not enough evidence to call it either way. This is not a middle verdict between harmful and harmless: it means unknown, and most changes ever catalogued sit here. A change can move out of this tier in either direction as evidence accumulates.ClinVar’s definitions
Likely benign: The submitter thinks the change does not cause the condition, with evidence that stops short of certain.ClinVar’s definitions
Benign: The submitter concluded the change does not cause the condition named next to it. Benign is a conclusion about that condition, not a clean bill of health about everything else.ClinVar’s definitions
Conflicting classifications: Submitters reached different conclusions about the same change. ClinVar records all of them rather than choosing one, which is why every submission is listed separately below with the date and the evidence level each carries.ClinVar’s definitions
Drug response: The change was recorded for its bearing on how someone responds to a particular medicine — dose, effectiveness or side effects — rather than for causing a disease.ClinVar’s definitions
Risk factor: The submitter recorded the change as raising the chance of a condition rather than causing it on its own.ClinVar’s definitions
Association: The change turned up more often in people with the condition than in people without it, in a study. A pattern across groups is not a cause in one person.ClinVar’s definitions
Protective: The submitter recorded the change as lowering the chance of a condition.ClinVar’s definitions
Affects: The change was linked to a trait that is not regarded as a disease — eye colour, or how a lab test reads.ClinVar’s definitions
Not provided: The submitter deposited a record without a conclusion of their own. The record is still here because it may carry conditions, citations or evidence.ClinVar’s definitions
Review status: How much checking sits behind a submission: whether the submitter followed published criteria, how many submitters there were, and whether they agreed. ClinVar shows it as a star rating from zero to four.Four stars, a practice guideline. Three, an expert panel. Two, several submitters using stated criteria and agreeing. One, a single submitter with stated criteria, or submitters who disagree. Zero, no criteria stated.ClinVar review status
Practice guideline: ClinVar’s highest evidence level, four stars: the classification comes from a published clinical practice guideline.ClinVar review status
Reviewed by expert panel: Three stars: a panel recognised by ClinVar for that gene or disease reviewed the evidence and issued one classification.ClinVar review status
Multiple submitters, no conflicts: Two stars: more than one submitter stated the criteria they used, and they agreed with each other.ClinVar review status
Conflicting classifications: One star: submitters stated their criteria and reached different conclusions. The disagreement is shown rather than resolved.ClinVar review status
Single submitter, criteria provided: One star: one submitter, who stated the criteria they judged by. Nobody else has deposited a view.ClinVar review status
No assertion criteria provided: Zero stars: the submitter did not say what standard they judged by. The conclusion may still be right; there is simply no way to weigh it.ClinVar review status

How common the change is

Counts from a large collection of people. How often something turns up is evidence about the change; it is not a conclusion about it.

Allele frequency: How common this spelling is in a large collection of people — the share of all the copies checked that carry it. Everyone carries two copies of most of the genome, so 0.01 means one copy in a hundred, not one person in a hundred.A change common in a population is, as a rule, an unlikely cause of a rare severe disease. Rarity on its own says nothing: most rare changes are harmless too, and frequency varies between ancestry groups.gnomAD Help/FAQ
gnomAD: The Genome Aggregation Database: DNA from many tens of thousands of people, pooled to show how common each change is. The people in it were not selected for being healthy, and no one group is a stand-in for everybody.gnomAD
gnomAD genomes: The part of gnomAD where the whole genome was read. Fewer people than the exome set, but coverage everywhere rather than only in genes.gnomAD on dataset selection
gnomAD exomes: The part of gnomAD where only the protein-coding stretches were read — about 2% of the genome. Many more people, but nothing outside those stretches.gnomAD on dataset selection

How the gene behaves as a whole

Population statistics about the gene the change falls in. They say nothing about any one change inside it.

Gene constraint: How badly a gene tolerates being broken, judged by counting how many damaging changes turn up in a large population against how many would be expected by chance. Far fewer than expected suggests carriers were selected against over evolutionary time. It describes the gene as a whole, not any one change in it.gnomAD on gene constraint
LOEUF: Observed against expected gene-breaking changes, taken at the cautious end of the estimate. It is the constraint number gnomAD now leads with.Low means intolerant. gnomAD suggests below 0.6 for picking out constrained genes, and below 0.35 for the strictest set. Around 1 means gene-breaking changes turn up about as often as chance would predict.gnomAD on gene constraint
pLI: The estimated probability that a gene cannot spare one of its two working copies. It runs from 0 to 1.gnomAD treats 0.9 and above as intolerant. It saturates — genes pile up at exactly 0 or 1 — and it is unreliable in short genes, which is why LOEUF is now generally preferred.gnomAD on gene constraint
o/e LoF: The plain ratio of gene-breaking changes seen in the population to the number expected by chance, without the cautious margin LOEUF adds.1 means as many as expected. 0.2 means a fifth as many, so such changes look to have been selected against. A small gene gives a noisy ratio, which is what LOEUF’s margin exists to handle.gnomAD on gene constraint
Missense z: How far the count of amino-acid-swapping changes in a gene departs from the count expected by chance, measured in standard deviations.Above about 3, far fewer than expected were seen, which suggests the gene tolerates such swaps poorly. Around 0 is as expected. Below 0, more than expected.gnomAD on gene constraint

Published work

What has been written about a change, and how it is counted.

Europe PMC: A free search across life-science papers, run by EMBL-EBI. It covers everything in PubMed and more besides, including preprints.About Europe PMC
Cited by: How many other indexed papers cite this one. It is a rough measure of how much notice a paper has drawn, not of whether it is right, and it favours older papers, which have had longer to be cited.About Europe PMC

What these definitions are not

They define words, not results. Nothing here is a conclusion about a variant you looked up, and nothing here is medical advice. Where a definition gives a threshold — pLI at 0.9, LOEUF at 0.35 — that threshold is the registry’s own suggestion for reading its own number, linked beside it, and not a rule this tool invented. If you are reading your own genetic data, take it to a licensed clinician or a certified genetic counsellor.