A patient with a rare genetic disorder has a missense variant in a disease-relevant gene. The SIFT algorithm predicts the variant is 'damaging' with 95% confidence. A clinician asks whether this variant is pathogenic. What is the correct response?
AYes — SIFT confidently predicts pathogenicity, so the variant should be classified as pathogenic
BNot necessarily — SIFT provides supporting computational evidence but pathogenicity requires integration with population frequency, conservation, functional data, and family segregation
CYes — a damaging missense variant in a relevant gene is pathogenic by definition under ACMG criteria
DNo — computational predictions like SIFT are unreliable and should not be used in clinical interpretation
SIFT and PolyPhen-2 provide computational supporting evidence — one line of evidence among many — not standalone pathogenicity determinations. ACMG classification requires weighing population frequency (is the variant common in gnomAD?), evolutionary conservation, experimental functional data, family segregation, and phenotype match, among other criteria. A SIFT 'damaging' call can be provided by perfectly benign variants at positions that happen to be conserved for structural reasons unrelated to the disease. Many SIFT-predicted-damaging variants are found in healthy individuals, confirming they are benign.
Question 2 Multiple Choice
Why does population frequency serve as the most powerful initial filter when interpreting a variant's pathogenicity for a rare Mendelian disease?
ABecause rare variants undergo stronger purifying selection and are therefore more likely to be harmful
BBecause a variant present at high frequency in healthy people cannot be causing a rare disease — disease prevalence logic makes it mathematically impossible
CBecause population databases like gnomAD include functional studies of each variant they list
DBecause common variants are tagged by GWAS studies, allowing direct disease association testing
The logic is straightforward: if a variant causes a rare disease (say, 1 in 100,000 individuals), it cannot be present in 5% of healthy individuals. Rare Mendelian diseases have population frequencies far below 1%, so any variant found at >0.1–1% in gnomAD healthy controls is almost certainly benign — regardless of what computational tools predict. This is why population frequency is the first filter applied and why finding a variant in gnomAD at appreciable frequency is considered strong evidence of benignity (ACMG criterion BA1/BS1).
Question 3 True / False
A variant classified as a VUS (variant of uncertain significance) may be reclassified as pathogenic or benign over time as additional evidence accumulates from more cases and functional studies.
TTrue
FFalse
Answer: True
VUS classification reflects current evidence insufficiency, not a permanent property of the variant. As more patients are sequenced, the variant may be observed multiple times in affected individuals with the same phenotype (increasing pathogenicity evidence) or found in healthy individuals (increasing benignity evidence). Functional experiments can demonstrate whether the variant disrupts protein function. The ACMG framework explicitly accommodates reclassification as evidence evolves, and large-scale reclassification projects have moved thousands of VUS to more definitive categories. VUS does not mean 'unknown forever' — it means 'we don't have enough evidence yet.'
Question 4 True / False
A de novo variant (not present in either parent) found in a patient with a matching disease phenotype should automatically be classified as pathogenic under ACMG guidelines.
TTrue
FFalse
Answer: False
De novo status in a disease-relevant gene with a matching phenotype is strong evidence for pathogenicity (ACMG criterion PS2), but it does not automatically yield a 'pathogenic' classification. The framework requires integration of multiple lines of evidence using a weighted point system. A de novo missense variant in a gene where only loss-of-function variants cause disease might still be classified as VUS. Additionally, some de novo variants occur by chance in genes that happen to be expressed in the brain without causing the observed phenotype. Automatic classification would skip the critical integration step that protects against misinterpretation.
Question 5 Short Answer
Why is a very rare variant not automatically pathogenic, and what additional lines of evidence are required before classifying it as pathogenic?
Think about your answer, then reveal below.
Model answer: Rarity is necessary but not sufficient for pathogenicity. Most rare variants in the human genome are simply rare neutral variants that never rose to common frequency by chance — the vast majority of sequence differences between individuals are benign. To classify a rare variant as pathogenic, interpreters look for: (1) functional evidence that it disrupts protein activity, splicing, or regulatory function; (2) evolutionary conservation suggesting the position is intolerant of change; (3) prior literature or ClinVar reports linking the same variant or gene to the disease; (4) family segregation showing the variant tracks with disease across generations; and (5) phenotype match confirming the patient's presentation is consistent with variants in this gene.
The ACMG framework assigns weighted criteria across these evidence types — strong, moderate, and supporting — and requires sufficient cumulative evidence before crossing the threshold to 'likely pathogenic' or 'pathogenic.' This system reflects the clinical reality that misclassification has direct consequences: incorrectly labeling a benign variant as pathogenic can lead to unnecessary surveillance, surgeries, or distress, while missing a true pathogenic variant delays diagnosis and treatment.