Glossary

Plain-language definitions of the clinical-genetics terms used across the portal.

Constraint

pLI
pLI (probability of loss-of-function intolerance) is gnomAD’s probability that a gene is intolerant to heterozygous loss-of-function variants.
pLI ≥ 0.9 flags a gene where a single inactivating variant may be disease-causing (haploinsufficiency).
LOEUF
LOEUF (Loss-of-function Observed/Expected Upper bound Fraction) is the upper bound of the 90% CI for the LoF observed/expected ratio — the preferred gnomAD v4 constraint metric. Lower = more intolerant to loss of function.
LOEUF < 0.35 marks a highly LoF-constrained gene.
Missense Z-score
How many standard deviations fewer missense variants are observed than expected across populations. Higher = more missense-constrained.
Z > 3.09 (p < 0.001) means the gene does not tolerate missense variation; OE missense < 0.6 is also considered constrained.
OE (LoF)
The observed/expected ratio of loss-of-function variants in the population — how many LoF variants are seen versus expected by chance.
The upper CI bound (LOEUF) ≤ 0.35 indicates strong LoF constraint.
OE (missense)
The observed/expected ratio of missense variants in the population.
OE ≤ 0.6 means fewer missense variants than expected — a missense-constraint signal.
pRec
pRec is gnomAD’s probability that a gene is intolerant to biallelic (recessive) loss of function while tolerating heterozygous loss of function.
pRec > 0.9 suggests heterozygous carriers are unaffected but biallelic loss is likely pathogenic.
Synonymous Z-score
A control metric — synonymous variants are largely neutral and expected near observed/expected = 1.0.
Large deviation usually points to annotation or coverage artifacts rather than biology.

Classification

VUS
Variant of Uncertain Significance — the evidence is insufficient to classify the variant as pathogenic or benign.
A VUS should not drive clinical decisions; reassess as new evidence emerges.
Pathogenic
A variant with strong evidence that it causes disease, per ACMG/AMP criteria.
Likely pathogenic
A variant with greater than ~90% certainty of being disease-causing, but short of the evidence required for Pathogenic.
Generally actionable like Pathogenic in most clinical contexts.
Benign / Likely benign
Benign means strong evidence the variant does not cause disease; Likely benign means greater than ~90% certainty it is harmless.
Conflicting interpretations
Different submitters have classified the variant differently (e.g. one Pathogenic, another VUS) — ClinVar’s “conflicting interpretations of pathogenicity”.
Inspect the individual submissions and their review status; weight expert-panel and recent assertions.
Review status (stars)
ClinVar’s 0–4 gold-star review status reflects how much independent support a classification has — from no assertion criteria up to expert-panel/practice-guideline review.
≥ 2 stars (multiple concordant submitters or an expert panel) is a stronger basis for clinical use.

Mechanism

Loss of function (LoF)
The variant reduces or abolishes the gene product’s normal activity — e.g. via a premature stop, frameshift, or deletion.
Gain of function (GoF)
The variant confers a new, increased, or constitutive activity rather than removing function.
Mechanism is variant-specific and needs functional/expert evidence; it is not inferable from gene-level data alone.
Dominant negative
The variant product interferes with the normal allele’s product (e.g. in multimeric proteins), causing more harm than simply losing one allele.
Haploinsufficiency
A single functional copy of the gene is not enough for normal function, so loss of one allele causes disease.
Suggested by high pLI / low LOEUF and ClinGen dosage haploinsufficiency scores.
Triplosensitivity
Having an extra copy (duplication) of the gene or region is itself harmful.
Assessed by ClinGen dosage triplosensitivity scores.

Population genetics

Allele frequency (AF)
The fraction of chromosomes in a reference population (e.g. gnomAD) that carry the variant.
High population AF argues against a highly penetrant disease cause (ACMG BA1/BS1).
Popmax
The highest allele frequency across the individual gnomAD ancestry groups, rather than the global average.
Used instead of global AF so a variant common in one ancestry is not missed.
faf95
The filtering allele frequency — the 95% confidence lower bound of the popmax AF. A conservative “is this too common to be pathogenic?” threshold.
faf95 above a disease’s maximum credible AF supports a benign classification (BS1/BA1).

ACMG criteria

PVS1
ACMG “pathogenic very strong”: a null variant (nonsense, frameshift, canonical splice, etc.) in a gene where loss of function is a known disease mechanism.
PM2
ACMG “pathogenic moderate” (commonly applied as Supporting): the variant is absent or extremely rare in population databases.
PP3 / BP4
ACMG computational evidence: in-silico predictors agree the variant is damaging (PP3, supports pathogenic) or benign (BP4, supports benign).
BA1 / BS1
ACMG population evidence for benign: the allele frequency is too high for a disease cause — stand-alone (BA1) or strong (BS1).

Inheritance

De novo
A variant present in the patient but in neither parent (newly arisen). Confirmed de novo status is ACMG evidence for pathogenicity (PS2/PM6).
Penetrance
The proportion of people carrying a pathogenic genotype who actually show the phenotype. Reduced penetrance means some carriers are unaffected.