# ACMG/AMP Criteria for Loss-of-Function Variants: How to Apply PVS1 and Its Modifications ## Key Takeaways - The PVS1 criterion in the ACMG/AMP framework is applied to predicted loss-of-function variants (nonsense, frameshift, canonical splice-site) when loss of function is the established disease mechanism. - PVS1 strength is modified based on variant type, transcript location, and evidence for a true null effect, with downgrades to strong or moderate strength necessary if nonsense-mediated decay is unlikely (e.g., variants in the last exon). - Canonical splice-site variants at positions +1, +2, -1, and -2 qualify for PVS1 if they disrupt splicing and lead to a frameshift or nonsense-mediated decay; in-frame exon skipping variants warrant downgrading. - Gene-specific specifications developed by Variant Curation Expert Panels refine general PVS1 recommendations, accounting for unique gene features, transcript structures, and disease mechanisms, which must be consulted by curators. - RNA assay data can provide experimental evidence to support or refine PVS1 strength, with PVS1_Strength proposed for capturing loss-of-function transcripts, while PS3/BS3 are reserved for functional assays not directly captured by splicing analysis. --- Variant curators face a recurring decision when interpreting nonsense, frameshift, and canonical splice-site variants: whether to assign the PVS1 criterion and at what strength. The 2015 ACMG/AMP framework defined PVS1 as a null variant in a gene where loss of function is the known mechanism of disease, but it left detailed application to expert judgment. The ClinGen Sequence Variant Interpretation Workgroup later published refined recommendations that specify how variant type, transcript location, and evidence for a true null effect should modify PVS1 strength. This article explains the PVS1 decision pathway, the conditions that justify downgrading from very strong to strong or moderate, and the gene-specific adjustments that Variant Curation Expert Panels have implemented. The practical outcome for readers is a reproducible workflow for applying PVS1 across different loss-of-function variant classes, with clear records and escalation criteria for ambiguous cases. ## The Problem PVS1 Solves in Variant Classification The ACMG/AMP classification system assigns pathogenicity evidence codes to sequence variants based on defined criteria. PVS1 is one of the pathogenic evidence codes and applies to predicted loss-of-function variants such as nonsense, frameshift, and splice-site changes. The original guideline stated that PVS1 should be used for a null variant in a gene where loss of function is the mechanism of disease, but it did not elaborate on how different types of loss-of-function variants should be weighted or how variant location should influence the code [1]. This gap created inconsistency across laboratories and expert panels. The ClinGen Sequence Variant Interpretation Workgroup addressed this by publishing detailed recommendations for applying PVS1. Their guidance covers the decision-making pathway for different loss-of-function variant types, the location of the variant within the transcript, and additional evidence that supports or refutes a true null effect [1]. The recommendations also introduced strength levels for PVS1, allowing curators to apply the code at very strong, strong, or moderate levels depending on the confidence that the variant truly abolishes gene function. The practical problem for curators is that PVS1 is not a single binary decision. A truncating variant in the last exon may escape nonsense-mediated decay and produce a truncated protein that retains partial function, which means PVS1 should not be applied at full strength. Similarly, a splice variant that does not alter the reading frame may produce an in-frame deletion instead of a null allele. The refined PVS1 framework provides explicit rules for these scenarios. ## Core Principles of the PVS1 Decision Pathway The ClinGen PVS1 recommendations organize the decision process around three questions. First, does the variant type predict a null effect? Second, is the variant located in a region where the predicted null effect is likely to occur? Third, is there additional evidence that supports or contradicts a true loss-of-function outcome? ### Variant Types That Qualify for PVS1 The PVS1 framework applies to nonsense variants, frameshift variants, canonical splice-site variants, initiation codon variants, and single-exon or multiexon deletions. Each variant type has specific conditions for PVS1 application. Nonsense and frameshift variants qualify when they are predicted to trigger nonsense-mediated decay, which requires the variant to be located upstream of the final exon-exon junction. Canonical splice-site variants at positions +1, +2, -1, and -2 qualify when they are predicted to disrupt splicing and lead to a frameshift or nonsense-mediated decay. Initiation codon variants qualify when they prevent translation initiation. The key distinction is whether the variant produces a complete absence of gene product or a truncated product that may retain function. The ClinGen recommendations emphasize that PVS1 should be applied at very strong strength only when the evidence supports a true null effect [1]. When the variant is located in the last exon or within 50 nucleotides of the final exon-exon junction, nonsense-mediated decay may not occur, and the resulting truncated protein may have residual function. In these cases, PVS1 should be downgraded. ### The Nonsense-Mediated Decay Boundary Nonsense-mediated decay is the cellular process that degrades mRNAs containing premature termination codons. The efficiency of this process depends on the position of the premature stop codon relative to the final exon-exon junction. Variants located in the last exon or near the end of the penultimate exon may escape nonsense-mediated decay and produce a truncated protein. The ClinGen recommendations specify that PVS1 should be applied at very strong strength for nonsense and frameshift variants predicted to undergo nonsense-mediated decay. For variants in the last exon or in the last 50 nucleotides of the penultimate exon, the recommendations advise downgrading PVS1 to strong or moderate, depending on the mechanism of disease and the predicted impact of the truncated protein [1]. This adjustment requires curators to know the exon structure of the relevant transcript and the position of the variant relative to the final exon-exon junction. ### Mechanism of Disease Determines PVS1 Applicability PVS1 is only applicable when loss of function is the established mechanism of disease for the gene. The ClinGen recommendations state that PVS1 should not be applied when the mechanism of disease is dominant-negative, gain-of-function, or when haploinsufficiency is not established [1]. This determination requires gene-specific knowledge that is often codified in ClinGen Variant Curation Expert Panel specifications. For genes where haploinsufficiency is the disease mechanism, a null variant is expected to cause disease by reducing gene dosage. For genes where a dominant-negative mechanism operates, a truncated protein may interfere with the normal protein product, and the variant may be pathogenic through a mechanism that is not captured by PVS1. In these cases, other evidence codes such as PS1 or PS3 may be more appropriate. ## The ClinGen PVS1 Strength Levels The ClinGen recommendations introduced a tiered system for PVS1 strength. The default application is PVS1 at very strong strength, which corresponds to the original ACMG/AMP designation. The recommendations define conditions for downgrading to strong (PVS1_Strong) or moderate (PVS1_Moderate) based on the confidence in a true null effect. ### PVS1 at Very Strong Strength PVS1 at very strong strength applies when the variant type and location provide high confidence in a null effect. This includes nonsense and frameshift variants predicted to undergo nonsense-mediated decay, canonical splice-site variants predicted to disrupt the reading frame, and initiation codon variants. The variant must be in a gene where loss of function is the established mechanism of disease. The ClinGen evaluation of the refined criterion across seven disease-specific groups showed 89% agreement with the new recommendations for 56 loss-of-function variants, with the remaining discrepancies attributed to disease-specific refinements [1]. This agreement rate indicates that the PVS1 framework provides consistent results across different curation contexts when applied according to the specified rules. ### PVS1 at Strong Strength PVS1 at strong strength applies when there is some uncertainty about the null effect. This includes nonsense and frameshift variants in the last exon or in the last 50 nucleotides of the penultimate exon, where nonsense-mediated decay may not occur. The truncated protein may retain partial function, so the evidence for a null effect is weaker. The decision to apply PVS1 at strong versus moderate strength depends on the predicted impact of the truncated protein. If the truncation removes a functionally important domain, the variant may still be expected to cause loss of function. If the truncation removes only a small portion of the protein and the remaining sequence retains known functional domains, the evidence for a null effect is weaker. ### PVS1 at Moderate Strength PVS1 at moderate strength applies when the evidence for a null effect is further reduced. This may occur for variants in the last exon that produce a truncated protein with substantial retained function, or for splice variants where the effect on the reading frame is uncertain. The ClinGen recommendations also address the use of PVS1_Strength to capture splicing assay data that provide experimental evidence for variants resulting in RNA transcripts with loss of function [2]. The 2023 ClinGen Splicing Subgroup recommendations propose repurposing the PVS1_Strength code to capture RNA assay evidence. When RNA analysis demonstrates that a variant results in transcripts with loss of function, this experimental evidence can support PVS1 at an appropriate strength level [2]. Conversely, RNA results demonstrating no splicing impact for intronic and synonymous variants may support the use of BP7 [2]. ## Gene-Specific PVS1 Specifications The ClinGen Variant Curation Expert Panels develop gene-specific specifications that refine the general PVS1 recommendations. These specifications account for the unique features of each gene, including transcript structure, protein domains, and the established mechanism of disease. ### How Variant Curation Expert Panels Modify PVS1 A 2024 evaluation of seven Variant Curation Expert Panels examined their loss-of-function criteria and how they applied PVS1 in pilot variant classifications [3]. The study found that VCEPs develop detailed definitions of what constitutes appropriate loss-of-function evidence, including specifications for variant type, location, nonsense-mediated decay boundaries, and additional evidence pointing to a true null effect [3]. The evaluation revealed that VCEPs vary in their approach to PVS1 strength levels and the conditions for downgrading. Some panels specify that PVS1 should be downgraded for variants in the last exon, while others require additional functional evidence before applying PVS1 at any strength. These gene-specific specifications are important for curators to consult before applying PVS1. ### Examples of Gene-Specific Adjustments For genes where the disease mechanism is haploinsufficiency, VCEPs typically apply PVS1 at very strong strength for nonsense and frameshift variants predicted to undergo nonsense-mediated decay. For genes where the mechanism is dominant-negative, VCEPs may specify that PVS1 should not be applied or should be downgraded because a truncated protein may exert a toxic effect. The ClinGen Hearing Loss Gene Curation Expert Panel provides an example of a gene-specific PVS1 specification. The GenOtoScope tool, which automates ACMG classification for hearing loss variants, implements the refined PVS1 criterion according to the Hearing Loss VCEP specifications [4]. This implementation demonstrates how gene-specific PVS1 rules can be codified in automated analysis tools. ## Splicing Variants and PVS1 Splice-site variants present a particular challenge for PVS1 application because the effect on the transcript must be predicted or experimentally determined. The ClinGen Splicing Subgroup published recommendations for integrating splicing evidence into the ACMG/AMP framework, including specific guidance for PVS1. ### Predicted Splicing Effects The Splicing Subgroup recommendations address how to apply PVS1 for variants that are predicted to affect splicing. The recommendations distinguish between variants at canonical splice sites, which are strong candidates for PVS1, and variants in other intronic or exonic regions that may create or disrupt splice sites. For canonical splice-site variants, the predicted effect on the reading frame determines PVS1 strength. If the variant is predicted to cause exon skipping that maintains the reading frame, the resulting in-frame deletion may not be a null allele, and PVS1 should be downgraded. If the variant is predicted to cause a frameshift or nonsense-mediated decay, PVS1 at very strong strength may be appropriate. ### RNA Assay Evidence The Splicing Subgroup recommendations propose using PVS1_Strength to capture RNA assay data that provide experimental evidence for variants resulting in RNA transcripts with loss of function [2]. This means that RNA analysis can support or refine the PVS1 code. For example, if RNA analysis demonstrates that a splice variant produces only transcripts that undergo nonsense-mediated decay, this evidence supports PVS1 at very strong strength. Conversely, RNA results demonstrating no splicing impact for intronic and synonymous variants may support the use of BP7 [2]. The recommendations also propose that PS3 and BS3 codes should be applied only for well-established assays that measure functional impact not directly captured by RNA-splicing assays [2]. This distinction helps curators avoid double-counting evidence. ### Calibrating Splice Prediction Tools The Splicing Subgroup recommendations include methodology for calibrating splice prediction tools [2]. This calibration process involves comparing computational predictions against empirically derived splicing evidence to determine appropriate thresholds for each tool. The calibrated tools can then be used to support PVS1 application for splice variants. The recommendations emphasize that splice prediction tools should be calibrated for the specific gene and variant context. A tool that performs well for one gene may not perform well for another, and the calibration process should account for this variability [2]. ## At a Glance: PVS1 Decision Table | Variant Type | Location | Predicted Effect | PVS1 Strength | Additional Evidence Needed | |---|---|---|---|---| | Nonsense or frameshift | Upstream of final exon-exon junction | Nonsense-mediated decay | Very strong | Gene has established loss-of-function mechanism | | Nonsense or frameshift | Last exon or last 50 nucleotides of penultimate exon | Truncated protein, possible escape from nonsense-mediated decay | Strong or moderate | Functional evidence for impact of truncated protein | | Canonical splice site (+1, +2, -1, -2) | Any intron | Frameshift or nonsense-mediated decay | Very strong | RNA assay confirmation recommended | | Canonical splice site | Any intron | In-frame exon skipping | Moderate or not applicable | RNA assay to determine reading frame | | Initiation codon | Exon 1 | No translation initiation | Very strong | Gene has established loss-of-function mechanism | | Single or multiexon deletion | Any region | Complete loss of gene product | Very strong | Confirmation of deletion breakpoints | ## Practical Workflow for Applying PVS1 The following workflow provides a step-by-step approach for applying PVS1 in variant classification. This workflow is designed for laboratory professionals and researchers who curate variants as part of diagnostic or research activities. ### Step 1: Confirm the Gene Has an Established Loss-of-Function Mechanism Before applying PVS1, confirm that loss of function is the established mechanism of disease for the gene. This information is typically available from ClinGen Variant Curation Expert Panel specifications, gene-specific databases, or the published literature. If the mechanism is dominant-negative or gain-of-function, PVS1 should not be applied. ### Step 2: Determine the Variant Type Identify the variant type from the sequencing data. Nonsense variants introduce a premature termination codon. Frameshift variants alter the reading frame. Canonical splice-site variants affect the +1, +2, -1, or -2 positions of an intron. Initiation codon variants affect the start codon. Each variant type has specific conditions for PVS1 application. ### Step 3: Determine the Variant Location Relative to the Final Exon-Exon Junction For nonsense and frameshift variants, determine whether the variant is located upstream of the final exon-exon junction. Variants in the last exon or in the last 50 nucleotides of the penultimate exon may escape nonsense-mediated decay. This determination requires knowledge of the transcript structure, including the exon boundaries and the position of the variant. ### Step 4: Assess the Predicted Effect on the Protein For variants that may escape nonsense-mediated decay, assess the predicted effect on the protein. Consider whether the truncated protein retains known functional domains and whether the truncation is expected to abolish protein function. This assessment may require consultation of protein domain databases and the published literature. ### Step 5: Apply PVS1 at the Appropriate Strength Apply PVS1 at very strong strength for variants with high confidence in a null effect. Downgrade to strong or moderate for variants with uncertainty about the null effect. Document the rationale for the strength level in the variant interpretation record. ### Step 6: Consider Additional Evidence Consider whether additional evidence supports or refutes a true null effect. RNA assay data can provide experimental evidence for the effect of splice variants on the transcript [2]. Functional assays may provide evidence for the impact of truncated proteins. This evidence can support the PVS1 strength level or may suggest that PVS1 should not be applied. ### Step 7: Document the Decision Record the PVS1 application in the variant interpretation file, including the variant type, location, predicted effect, strength level, and the evidence supporting the decision. This documentation is essential for reproducibility and for comparison across laboratories. ## Records and Measurements for PVS1 Application Accurate record-keeping is essential for consistent PVS1 application. The following records should be maintained for each variant classified. ### Variant Information Record the genomic position, reference and alternate alleles, and the variant type. Include the transcript identifier used for the interpretation, as different transcripts may have different exon structures and nonsense-mediated decay boundaries. ### Transcript Information Record the exon structure of the relevant transcript, including the position of the final exon-exon junction. This information is necessary for determining whether a variant is predicted to undergo nonsense-mediated decay. ### PVS1 Decision Record the PVS1 strength level applied and the rationale for the decision. Include the variant type, location, predicted effect, and any additional evidence considered. ### Supporting Evidence Record any RNA assay data, functional assay data, or computational predictions that support the PVS1 application. Include the source of the evidence and the date the evidence was generated. ### Quality Control Measures Record the quality control measures applied to the variant call, including sequencing depth, genotype quality, and variant caller confidence. These measures ensure that the variant being interpreted is a true variant and not a sequencing artifact. ## Common Failure Patterns in PVS1 Application Several recurring errors occur in PVS1 application. Recognizing these patterns can help curators avoid them. ### Applying PVS1 Without Confirming the Loss-of-Function Mechanism A common error is applying PVS1 to a gene where loss of function is not the established mechanism of disease. This error can lead to overclassification of variants as pathogenic. Curators should confirm the mechanism of disease before applying PVS1. ### Ignoring the Nonsense-Mediated Decay Boundary Another common error is applying PVS1 at very strong strength for variants in the last exon or near the final exon-exon junction. These variants may escape nonsense-mediated decay and produce a truncated protein with residual function. The ClinGen recommendations specify that PVS1 should be downgraded in these cases [1]. ### Applying PVS1 to Splice Variants Without Assessing the Reading Frame Splice variants that cause in-frame exon skipping may produce a protein with an internal deletion instead of a null allele. Applying PVS1 at very strong strength to these variants is inappropriate. RNA assay data can help determine the effect on the reading frame [2]. ### Failing to Consult Gene-Specific Specifications Variant Curation Expert Panels develop gene-specific specifications that may modify the general PVS1 recommendations. Failing to consult these specifications can lead to inconsistent PVS1 application. The 2024 evaluation of VCEPs found that panels vary in their approach to PVS1 strength levels and downgrading conditions [3]. ### Double-Counting Evidence The Splicing Subgroup recommendations distinguish between PVS1 and PS3/BS3 evidence. PVS1 captures predicted or observed loss-of-function effects on the transcript, while PS3 and BS3 capture functional assay results [2]. Applying both codes for the same evidence can lead to overcounting. ## Limitations of the PVS1 Framework The PVS1 framework has several limitations that curators should recognize. ### Dependence on Transcript Annotation PVS1 application depends on accurate transcript annotation, including exon boundaries and the position of the final exon-exon junction. Different transcripts may have different structures, and the choice of transcript can affect the PVS1 decision. Curators should use the transcript that is most relevant to the disease mechanism. ### Uncertainty in Nonsense-Mediated Decay Prediction The prediction of nonsense-mediated decay is based on the position of the premature termination codon relative to the final exon-exon junction. However, nonsense-mediated decay efficiency can vary between genes and cell types. The ClinGen recommendations acknowledge this uncertainty by providing strength levels instead of a binary decision [1]. ### Limited Experimental Validation For many variants, no RNA or functional assay data are available. In these cases, PVS1 application relies on computational predictions and the variant type and location. The ClinGen recommendations provide a framework for these situations, but the evidence for a null effect remains indirect. ### Gene-Specific Variability The 2024 evaluation of VCEPs found that panels vary in their PVS1 specifications [3]. This variability means that the same variant may receive different PVS1 strength levels from different panels. Curators should be aware of this variability and consult the relevant gene-specific specifications. ## Professional Escalation Criteria Certain situations warrant escalation to a senior curator, clinical geneticist, or expert panel. The following criteria indicate when PVS1 application requires additional expertise. ### Ambiguous Mechanism of Disease If the mechanism of disease for the gene is not clearly established, escalate the case to a senior curator or clinical geneticist. Applying PVS1 without a confirmed loss-of-function mechanism can lead to misclassification. ### Variants in the Nonsense-Mediated Decay Boundary Region Variants in the last exon or in the last 50 nucleotides of the penultimate exon require careful assessment of the predicted impact of the truncated protein. If the functional impact is uncertain, escalate the case for expert review. ### Splice Variants With Uncertain Reading Frame Effects Splice variants where the effect on the reading frame is uncertain require RNA assay data or expert assessment. If RNA assay data are not available, escalate the case to determine whether PVS1 should be applied and at what strength. ### Discrepancies Between Computational Predictions and Experimental Data If computational predictions suggest a null effect but RNA or functional assay data suggest otherwise, escalate the case. The Splicing Subgroup recommendations provide guidance for integrating these different types of evidence [2]. ### Gene-Specific Specification Conflicts If the gene-specific VCEP specification conflicts with the general ClinGen PVS1 recommendations, escalate the case to determine which specification should take precedence. The VCEP specifications are developed for specific genes and diseases and may include refinements that are not captured in the general recommendations [3]. ## Bioinformatics Tools for PVS1 Application Several bioinformatics tools can support PVS1 application by automating parts of the decision process. ### GenOtoScope GenOtoScope is an open-source tool that automates the analysis of ACMG/AMP criteria that can be assessed without individual patient information, including the refined PVS1 criterion [4]. The tool provides a command-line application for batch classification and a web interface for single variant classification. GenOtoScope was developed for hearing loss variants and implements the ClinGen Hearing Loss Gene Curation Expert Panel specifications [4]. ### Variant Annotation Tools Variant annotation tools can provide the transcript information needed for PVS1 application, including exon boundaries, variant position, and predicted effect on the protein. These tools are available through the NCBI data resources [5] and other bioinformatics platforms. ### Splice Prediction Tools Splice prediction tools can support the assessment of splice variants. The Splicing Subgroup recommendations include methodology for calibrating these tools against empirically derived splicing evidence [2]. Calibrated tools can provide more reliable predictions for PVS1 application. ### Workflow Platforms Workflow platforms such as Galaxy and nf-core provide reproducible analysis pipelines that can support variant interpretation workflows. The Galaxy Training Network offers accessible workflow training and analysis tutorials [6], and the nf-core documentation provides community pipeline standards and usage guidance [7]. These platforms can help laboratories implement consistent variant interpretation workflows. ## Training and Education for PVS1 Application Proper PVS1 application requires training in variant interpretation and the ACMG/AMP framework. Several resources are available for this training. ### Bioinformatics Training Resources The EMBL-EBI Training program offers learning pathways for bioinformatics data resources and practical analysis education [8]. These resources can help laboratory professionals develop the computational skills needed for variant interpretation. ### Reproducible Analysis Training The Carpentries lessons provide foundational training in computing, data, shell, Git, and programming [9]. These skills are essential for implementing reproducible variant interpretation workflows. ### Bioconductor Resources The Bioconductor project provides official documentation for packages, workflows, installation, and reproducible genomic analysis [10]. These resources can support the development of custom variant interpretation pipelines. ### Galaxy Training The Galaxy Training Network offers accessible workflow training and analysis tutorials [6]. These resources can help laboratories implement reproducible variant interpretation workflows without requiring extensive programming skills. ## Quality Control in PVS1 Application Quality control measures are essential for consistent PVS1 application across variants and curators. ### Standardized Decision Documentation Use a standardized format for documenting PVS1 decisions, including the variant type, location, predicted effect, strength level, and supporting evidence. This standardization facilitates comparison across curators and laboratories. ### Regular Calibration Sessions Conduct regular calibration sessions where curators review PVS1 decisions for a set of variants and discuss discrepancies. These sessions can identify areas where the PVS1 framework is being applied inconsistently. ### Audit of PVS1 Decisions Periodically audit PVS1 decisions to identify patterns of overapplication or underapplication. The audit should include a review of variants where PVS1 was downgraded and the rationale for the downgrade. ### Integration With Gene-Specific Specifications Ensure that PVS1 decisions are consistent with the relevant gene-specific VCEP specifications. The 2024 evaluation of VCEPs found that panels vary in their PVS1 specifications, so curators should consult the relevant specifications before finalizing a decision [3]. ## Safety and Regulatory Context Variant interpretation has direct implications for patient care, and errors in PVS1 application can lead to incorrect clinical decisions. The following considerations are relevant for laboratories performing variant interpretation. ### Clinical Validation Requirements Laboratories performing clinical variant interpretation must validate their workflows and demonstrate that PVS1 is applied consistently and accurately. This validation should include a review of PVS1 decisions against the ClinGen recommendations and gene-specific specifications. ### Reporting Requirements Variant interpretation reports should include the PVS1 strength level and the rationale for the decision. This information allows clinicians and other laboratories to understand the basis for the classification. ### Data Sharing Sharing variant interpretation data through databases such as ClinVar can improve consistency across laboratories. The NCBI data resources provide access to these databases and the associated search systems [5]. ### Professional Oversight Variant interpretation should be performed under the oversight of qualified professionals, including clinical geneticists and molecular pathologists. Escalation criteria should be defined for cases that require expert review. ## Building a PVS1 Decision Log and Audit Trail for Consistent Curation Variant curators often apply PVS1 correctly in isolation but struggle to maintain consistency across a large cohort or over time. The gap between knowing the ClinGen PVS1 rules and applying them uniformly across hundreds of variants is a practical problem that a structured decision log can solve. A decision log is a working document that records the reasoning behind each PVS1 strength assignment, making the curation process transparent, reproducible, and auditable. This section provides a concrete framework for building such a log, including the fields to capture, the review triggers, and the common patterns that indicate the log is not being used effectively. ### Why a Decision Log Matters for PVS1 The ClinGen PVS1 recommendations were developed to reduce inconsistency across laboratories, and the evaluation of seven disease-specific groups showed 89% agreement with the new recommendations for 56 loss-of-function variants [1]. The remaining 11% of discrepancies were appropriately due to disease-specific refinements, which means that even with clear guidance, expert judgment plays a role in PVS1 application [1]. A decision log captures that expert judgment in a way that can be reviewed, challenged, and improved. Without a decision log, a curator may apply PVS1 at very strong strength for a nonsense variant in one gene and at strong strength for an identical variant type in another gene without being able to articulate why the strength levels differed. The log forces the curator to document the variant type, the transcript location, the predicted effect, and the evidence that supports the chosen strength level. This documentation is essential for internal quality control, for external audits, and for the ClinGen Variant Curation Expert Panel review process. ### Core Fields for the PVS1 Decision Log The decision log should capture enough information to reconstruct the full reasoning for each PVS1 assignment. The following fields provide a minimum standard for a useful log entry. #### Variant Identification Fields Record the genomic position using the reference genome build, the reference and alternate alleles, and the variant type. Include the transcript identifier used for the interpretation, because different transcripts may have different exon structures and nonsense-mediated decay boundaries. The choice of transcript can change the PVS1 strength assignment, so the log must record which transcript was used and why. #### PVS1 Decision Fields Record the PVS1 strength level applied, which can be very strong, strong, or moderate. Document the variant type, the location relative to the final exon-exon junction, and the predicted effect on the transcript or protein. For nonsense and frameshift variants, record whether the variant is predicted to undergo nonsense-mediated decay. For splice variants, record the predicted effect on the reading frame and whether RNA assay data are available. #### Evidence Fields Record the evidence that supports the PVS1 strength level. This includes the established mechanism of disease for the gene, which should be confirmed before PVS1 is applied. Include any RNA assay data, functional assay data, or computational predictions that support the decision. The ClinGen Splicing Subgroup recommends using PVS1_Strength to capture RNA assay data that demonstrate loss-of-function transcripts [2]. Record the source of the evidence and the date the evidence was generated. #### Review Fields Record the name of the curator who made the decision, the date of the decision, and the name of any reviewer who confirmed the decision. Include a field for the next review date, which is important for variants where new evidence may become available. ### Building the Decision Log in Practice The decision log can be maintained in a spreadsheet, a database, or a dedicated variant interpretation platform. The format matters less than the consistency of the fields and the discipline of completing every field for every variant. #### Step 1: Define the Log Structure Create a template with the core fields described above. Include dropdown menus for the PVS1 strength level, the variant type, and the predicted effect to reduce free-text variability. Add a free-text field for the rationale, but require that the rationale reference specific evidence instead of general statements. #### Step 2: Establish Entry Rules Define when a log entry is required. A log entry should be created for every variant where PVS1 is considered, including variants where PVS1 is considered but not applied. This captures the reasoning for not applying PVS1, which is as important as the reasoning for applying it. For example, if a curator considers PVS1 for a variant in a gene with a dominant-negative mechanism of disease and decides not to apply it, that decision should be logged. #### Step 3: Implement a Review Workflow Define who reviews the log entries and how often. A second curator should review PVS1 assignments at very strong strength, because these assignments have the greatest impact on the final variant classification. A senior curator should review all PVS1 assignments at strong or moderate strength where the variant is in the last exon or in the last 50 nucleotides of the penultimate exon, because these are the cases where the ClinGen recommendations allow the most discretion [1]. #### Step 4: Conduct Regular Audits Schedule a periodic audit of the decision log, such as quarterly or after every 100 variants. The audit should compare PVS1 assignments across similar variant types and genes to identify patterns of inconsistency. For example, if one curator consistently applies PVS1 at strong strength for last-exon variants while another curator applies moderate strength for the same variant type, the audit will surface this discrepancy for discussion. ### Using the Decision Log to Troubleshoot PVS1 Application The decision log is beyond a record-keeping tool. It is a troubleshooting instrument that can identify where the PVS1 framework is being applied inconsistently and where additional training or gene-specific guidance is needed. #### Pattern 1: Strength Level Drift If the audit reveals that PVS1 strength levels have drifted over time, such as a gradual shift from very strong to strong for the same variant type, this may indicate that curators are becoming more conservative or more permissive without a clear rationale. The decision log can identify when the drift began and which variants were affected. #### Pattern 2: Gene-Specific Inconsistency If the audit reveals that PVS1 is applied at different strength levels for the same variant type in different genes, this may indicate that curators are not consistently consulting the gene-specific Variant Curation Expert Panel specifications. The 2024 evaluation of seven VCEPs found that panels vary in their approach to PVS1 strength levels and downgrading conditions [3]. The decision log should record which VCEP specification was consulted for each gene. #### Pattern 3: Evidence Double-Counting If the audit reveals that PVS1 and PS3 or BS3 are being applied for the same evidence, this may indicate that curators are double-counting. The ClinGen Splicing Subgroup recommends that PS3 and BS3 codes be applied only for well-established assays that measure functional impact not directly captured by RNA-splicing assays [2]. The decision log should record the source of each evidence code so that double-counting can be identified. #### Pattern 4: Missing RNA Evidence If the audit reveals that splice variants are being classified without RNA assay data, this may indicate that curators are relying on computational predictions alone. The Splicing Subgroup recommendations provide a process for integrating splicing-related considerations when developing a gene-specific PVS1 decision tree [2]. The decision log should flag splice variants where RNA evidence is missing and a follow-up is needed. ### Records and Measurements for the Decision Log The decision log should be treated as a formal record that is subject to the same quality standards as the variant interpretation itself. The following measurements can be used to assess the quality of the decision log. #### Log Completion Rate Measure the percentage of variants where a log entry was created out of the total variants where PVS1 was considered. A completion rate below 95% indicates that the log is not being used consistently. #### Review Turnaround Time Measure the time between the initial PVS1 assignment and the review by a second curator. A long turnaround time may indicate that reviews are not being prioritized. #### Audit Finding Rate Measure the number of audit findings per 100 variants reviewed. A high finding rate may indicate that the PVS1 framework is not being applied consistently, while a very low finding rate may indicate that the audit is not sufficiently rigorous. ### Common Failure Patterns in Decision Log Usage Recognizing the common ways that decision logs fail can help curators avoid these pitfalls. #### Treating the Log as Paperwork The most common failure is treating the decision log as administrative paperwork instead of as a scientific record. If the log entries are completed after the classification is finalized instead of during the curation process, the log will not capture the actual reasoning. The log should be completed at the time the PVS1 decision is made. #### Using Vague Rationale A log entry that states "PVS1 applied at very strong strength because the variant is a nonsense variant" does not capture the reasoning. The rationale should reference the specific evidence, such as the variant location relative to the final exon-exon junction, the established mechanism of disease for the gene, and any RNA assay data. #### Failing to Update the Log If new evidence becomes available, such as RNA assay data that changes the predicted effect of a splice variant, the log entry should be updated. The ClinGen Splicing Subgroup recommendations propose repurposing the PVS1_Strength code to capture splicing assay data that provide experimental evidence for variants resulting in RNA transcripts with loss of function [2]. This new evidence should be reflected in the log. #### Ignoring the Audit Findings An audit that identifies inconsistencies is only useful if the findings are acted upon. The audit should result in specific actions, such as retraining on the ClinGen PVS1 recommendations, consulting a gene-specific VCEP specification, or revising the decision log template. ### Professional Escalation Criteria for the Decision Log Certain situations warrant escalation beyond the routine review workflow. #### Repeated Inconsistency If the audit reveals the same inconsistency in multiple review cycles, escalate the issue to a senior curator or the laboratory director. This may indicate a training gap or a need for a revised gene-specific specification. #### Disagreement Between Curators If two curators cannot agree on the appropriate PVS1 strength level for a variant, escalate the case to a senior curator or to the relevant ClinGen Variant Curation Expert Panel. The 2024 evaluation of VCEPs found that panels vary in their PVS1 specifications, so the relevant panel should be consulted [3]. #### New Evidence That Changes the Classification If new RNA or functional evidence changes the PVS1 strength level for a variant that has already been reported, escalate the case to determine whether the classification should be revised and whether the change affects other variants in the same gene. ### Integration With Bioinformatics Workflows The decision log can be integrated with bioinformatics workflows to reduce manual data entry and improve consistency. Automated tools such as GenOtoScope can apply PVS1 according to gene-specific specifications and generate the variant information fields for the log [4]. The curator can then focus on the evidence fields and the rationale, which require expert judgment. Workflow platforms such as Galaxy and nf-core provide reproducible analysis pipelines that can support variant interpretation workflows [6][7]. These platforms can be configured to generate the variant information fields for the decision log automatically, reducing the risk of transcription errors. The NCBI data resources provide access to databases and search systems that can support the evidence fields in the decision log [5]. Curators can use these resources to confirm the established mechanism of disease for a gene and to identify relevant RNA or functional assay data. ### Training for Decision Log Usage The decision log is only as useful as the training that supports it. Curators should be trained on the purpose of the log, the fields that must be completed, and the review workflow. The EMBL-EBI Training program offers learning pathways for bioinformatics data resources and practical analysis education [8]. The Carpentries lessons provide foundational training in computing, data, shell, Git, and programming [9]. These resources can support the development of the computational skills needed to maintain a decision log effectively. The Bioconductor project provides official documentation for packages, workflows, installation, and reproducible genomic analysis [10]. These resources can support the development of custom tools for decision log analysis and audit reporting. ## Frequently Asked Questions ### What is the difference between PVS1 and PVS1_Strong? PVS1 at very strong strength applies when there is high confidence in a null effect, such as a nonsense or frameshift variant predicted to undergo nonsense-mediated decay in a gene with an established loss-of-function mechanism. PVS1_Strong applies when there is some uncertainty about the null effect, such as a variant in the last exon that may escape nonsense-mediated decay and produce a truncated protein with residual function [1]. ### When should PVS1 be downgraded to moderate strength? PVS1 should be downgraded to moderate strength when the evidence for a null effect is further reduced. This may occur for variants in the last exon that produce a truncated protein with substantial retained function, or for splice variants where the effect on the reading frame is uncertain. The ClinGen recommendations provide specific guidance for these scenarios [1]. ### Can PVS1 be applied to splice variants? Yes, PVS1 can be applied to canonical splice-site variants at positions +1, +2, -1, and -2. The strength of PVS1 depends on the predicted effect on the reading frame. If the variant is predicted to cause a frameshift or nonsense-mediated decay, PVS1 at very strong strength may be appropriate. If the variant is predicted to cause in-frame exon skipping, PVS1 should be downgraded or may not be applicable [2]. ### How does RNA assay data affect PVS1 application? RNA assay data can provide experimental evidence for the effect of a variant on the transcript. The ClinGen Splicing Subgroup recommends using PVS1_Strength to capture RNA assay data that demonstrate loss-of-function transcripts [2]. Conversely, RNA results demonstrating no splicing impact for intronic and synonymous variants may support the use of BP7 [2]. ### What is the role of Variant Curation Expert Panels in PVS1 application? Variant Curation Expert Panels develop gene-specific specifications that refine the general PVS1 recommendations. These specifications account for the unique features of each gene, including transcript structure, protein domains, and the established mechanism of disease. A 2024 evaluation found that VCEPs vary in their approach to PVS1 strength levels and downgrading conditions [3]. ### How do I determine if a gene has an established loss-of-function mechanism? The established mechanism of disease for a gene is typically documented in ClinGen Variant Curation Expert Panel specifications, gene-specific databases, or the published literature. If the mechanism is dominant-negative or gain-of-function, PVS1 should not be applied. If the mechanism is unclear, escalate the case to a senior curator or clinical geneticist. ### What is the nonsense-mediated decay boundary and why does it matter? The nonsense-mediated decay boundary is the position of the final exon-exon junction in a transcript. Nonsense and frameshift variants located upstream of this boundary are predicted to trigger nonsense-mediated decay, which degrades the mRNA and prevents the production of a truncated protein. Variants in the last exon or in the last 50 nucleotides of the penultimate exon may escape nonsense-mediated decay and produce a truncated protein with residual function [1]. ### Can automated tools apply PVS1 reliably? Automated tools such as GenOtoScope can apply PVS1 according to gene-specific specifications, but they require careful validation and calibration. GenOtoScope achieved the best average accuracy and precision compared with two other variant classification tools in an evaluation using hearing loss data sets [4]. However, automated tools should be used as a support for human curation, not as a replacement for expert judgment. ## Related Bioinformatics Guides - [Detecting Structural Variants with Long-Read Sequencing: Methods and Considerations](/knowledge/bioinformatics/detecting-structural-variants-with-long-read-sequencing-methods-and-considerations) - [How to Interpret Gene Set Enrichment Analysis Results](/knowledge/bioinformatics/how-to-interpret-gene-set-enrichment-analysis-results) - [Persistent Identifiers in Bioinformatics: Selection Criteria and Implementation Workflow](/knowledge/bioinformatics/persistent-identifiers-in-bioinformatics-selection-criteria-and-implementation-workflow) - [Functional Metagenomics: From Gene Prediction to Pathway Reconstruction](/knowledge/bioinformatics/functional-metagenomics-from-gene-prediction-to-pathway-reconstruction) - [RNA-Seq vs ChIP-Seq: Complementary Approaches for Gene Regulation](/knowledge/bioinformatics/rna-seq-vs-chip-seq-complementary-approaches-for-gene-regulation) ## Related Clinical & Scientific Guides * [A Practical Guide to Detecting Antimicrobial Resistance Genes in Shotgun Metagenomic Data](/knowledge/bioinformatics/a-practical-guide-to-detecting-antimicrobial-resistance-genes-in-shotgun-metagenomic-data) * [Computational Immunology: Modeling the Immune System](/knowledge/bioinformatics/computational-immunology-modeling-the-immune-system) * [How to Set Hard Filters for Germline Variant Calling: A Practical Guide to GATK Best Practices](/knowledge/bioinformatics/how-to-set-hard-filters-for-germline-variant-calling-a-practical-guide-to-gatk-best-practices) ## References and Further Reading [1] [Recommendations for interpreting the loss of function PVS1 ACMG/AMP variant criterion.](https://pubmed.ncbi.nlm.nih.gov/30192042). Human mutation, 2018. [2] [Using the ACMG/AMP framework to capture evidence related to predicted and observed impact on splicing: Recommendations from the ClinGen SVI Splicing Subgroup.](https://pubmed.ncbi.nlm.nih.gov/37352859). American journal of human genetics, 2023. [3] [Evaluating ClinGen variant curation expert panels' application of PVS1 code.](https://pubmed.ncbi.nlm.nih.gov/38199457). European journal of medical genetics, 2024. [4] [GenOtoScope: Towards automating ACMG classification of variants associated with congenital hearing loss.](https://pubmed.ncbi.nlm.nih.gov/36129964). PLoS computational biology, 2022. [5] [NCBI Data Resources](https://www.ncbi.nlm.nih.gov/). National Center for Biotechnology Information. [6] [Galaxy Training Network](https://training.galaxyproject.org/). Galaxy Project. [7] [nf-core Documentation](https://nf-co.re/docs). nf-core. [8] [EMBL-EBI Training](https://www.ebi.ac.uk/training). European Bioinformatics Institute. [9] [The Carpentries Lessons](https://carpentries.org/lessons). The Carpentries. [10] [Bioconductor](https://bioconductor.org/). Bioconductor Project. > This article is educational and does not replace validated analysis plans, institutional policy, clinical interpretation, or specialist review.