View Categories

PVS1 – Interpreting Loss-of-Function (LoF) Variants in SeqSMART

3 min read

Overview #

Loss-of-function (LoF) variants — such as nonsense, frameshift, canonical splice site, initiation codon, or multi-exon deletions — can lead to loss or severe reduction of gene product activity. The ACMG/AMP 2015 guidelines designate PVS1 (“Pathogenic Very Strong”) for such variants in genes where LoF is an established disease mechanism.
However, LoF interpretation requires context: not every premature stop or splice variant truly results in loss of function. Following ClinGen’s 2018 recommendations

PVS1

, SeqSMART integrates a structured decision-making process that accounts for variant type, transcript context, NMD prediction, exon relevance, and disease mechanism evidence.

SeqSMART’s Multi-Step PVS1 Evaluation Framework #

SeqSMART implements an automated yet transparent process that mirrors ClinGen’s PVS1 decision tree while incorporating additional data-driven refinements.

Step 1. Confirming LoF as a Disease Mechanism #

Before assigning PVS1, SeqSMART verifies that LoF is a known and validated mechanism for the gene-disease association.

  • We use ClinGen gene–disease validity classifications and constraint metrics (e.g., pLI scores, LoF intolerance) from gnomAD.
  • PVS1 is applied only if the gene has a Strong or Definitive association and multiple known pathogenic LoF variants across distinct exons.
  • For genes with Moderate evidence, the system downgrades strength to PVS1_Strong or PVS1_Moderate depending on LoF variant distribution and supporting evidence (e.g., knockout animal models).

If LoF is not a known disease mechanism, PVS1 is not applied at any level.

Step 2. Variant-Type–Specific Evaluation #

A. Nonsense and Frameshift Variants #

These variants are first assessed for nonsense-mediated decay (NMD) potential and the biological relevance of affected exons.

  1. NMD Prediction
    • PVS1 assumes NMD occurs when the premature termination codon (PTC) lies >50 bp upstream of the final exon–exon junction.
    • If NMD is predicted, PVS1 or PVS1_Strong is considered, depending on exon relevance and LoF mechanism strength.
  2. NMD-Unlikely Cases
    • If the PTC occurs within the last exon or the terminal 50 bp of the penultimate exon, NMD is unlikely.
    • SeqSMART then checks whether the truncated region includes critical domains (e.g., catalytic or DNA-binding motifs).
      • If critical: PVS1_Strong
      • If noncritical: PVS1_Moderate
  3. Alternative Transcripts
    • If the affected exon is absent from biologically relevant transcripts or tolerant to LoF (based on gnomAD frequency data), PVS1 is not applied.

B. Canonical ±1 or ±2 Splice Site Variants #

Canonical splice variants are typically LoF unless a rescue mechanism exists.

  • SeqSMART predicts outcomes via SpliceAI and in-house models:
    • Exon skipping, intron retention, or use of cryptic splice sites are evaluated.
    • If the exon length is divisible by three (potential in-frame skipping), strength may be reduced (e.g., to PVS1_Moderate).
  • Nearby cryptic sites (±20 bp) are assessed to determine whether normal splicing could be restored.
  • If normal splicing may be preserved, or the impact is uncertain, PVS1 is downgraded or not applied.

Note: PP3 (in silico prediction) is not used concurrently with PVS1 for splice variants to avoid double counting.

C. Start-Loss Variants #

Variants abolishing the canonical start codon (ATG) are analyzed for downstream reinitiation potential.

  • If alternative in-frame ATG codons are present early in the transcript, the variant may yield a partially functional protein → PVS1_Moderate or Supporting.
  • If no alternative start codon exists, and the gene’s disease mechanism depends on full-length translation → PVS1_Strong.

SeqSMART cross-references known pathogenic variants upstream of alternative start sites to calibrate strength, as recommended by ClinGen.

D. Splice Site and Exon Deletions #

For exon-level deletions, SeqSMART evaluates:

  • Frame impact — whether the deletion causes a frameshift.
  • NMD prediction for transcripts with frame disruptions.
  • Transcript inclusion — if the deleted exon is absent from biologically relevant isoforms, PVS1 is not applied.
  • Whole-gene deletions in haploinsufficient genes are considered pathogenic (PVS1 level, barring conflicting evidence).

E. Duplications #

If the duplicated region’s orientation or insertion site is unknown:

  • SeqSMART assumes tandem duplication by default (supported by >80% evidence from ClinGen).
  • If predicted to cause a frameshift and NMD: PVS1_Strong.
  • If duplication is in-frame or uncertain: downgraded to PVS1_Moderate or unassigned.

Step 3. Integrating Functional Domain and Transcript Data #

For every variant, SeqSMART integrates:

  • UniProt functional domain annotations
  • RefSeq/Ensembl transcript structures
  • ClinVar pathogenic variant distribution
  • gnomAD LoF frequency data

If the truncated or deleted region is functionally critical (supported by downstream pathogenic variants or domain loss), PVS1 may be elevated (e.g., from Strong → Very Strong).

Step 4. Assigning PVS1 Strength #

SeqSMART applies the following evidence-weighted levels consistent with the Bayesian calibration from ClinGen:

PVS1 StrengthTypical Scenario
Very Strong (PVS1)Confirmed NMD-triggering LoF in a gene with established LoF disease mechanism
Strong (PVS1_Strong)NMD-unlikely but critical domain truncated, or moderate LoF evidence
Moderate (PVS1_Moderate)In-frame or partial truncation with possible residual function
Supporting (PVS1_Supporting)Start-loss or uncertain LoF impact with suggestive evidence

To prevent double-counting, PM4 (in-frame deletion/stop-loss) is not used in combination with any PVS1 level.

Transparency and Manual Review #

Every PVS1 call in SeqSMART is fully traceable.
Users can review:

  • The transcript affected
  • NMD prediction results
  • Domain overlap
  • Population data and ClinVar references

Manual overrides are supported, allowing expert reviewers to adjust the PVS1 level when experimental or literature data warrant reassessment.

SeqSMART’s PVS1 framework operationalizes ClinGen’s recommendations into a scalable, gene-aware, and transcript-sensitive evaluation system.
By combining predictive modeling, curated databases, and configurable expert oversight, SeqSMART ensures that each loss-of-function variant is assessed with precision, transparency, and adherence to clinical best practices.