Clinical Significance Tab

The Clinical Significance tab summarizes essential variant-level and gene-level information, and indicates the gene's associated diseases. It provides crucial insights needed to evaluate a variant’s potential pathogenicity using AI-supported evidence and public databases.

This tab contains:

1. Variant Info

Provides key details about the variant, including:

• Variant type (SNV, CNV, Indel, etc.)

• Main effect (e.g., missense, nonsense, synonymous)

• Zygosity for each sequenced family member (Het, Hom, Hemi, Ref)

• Gene symbol and HGVS nomenclature for both coding DNA and protein change

• Transcript label:

  • Canonical: if it’s the primary reference transcript

  • Curate: if the transcript was chosen in your Curate database

  Users can change the reference transcript using the dropdown or add a custom one if needed. HGVS fields may be editable for certain variant types.

  For certain variants, such as upstream or downstream gene variants, HGVS descriptions may not be available. In these cases, you have the option to manually input coding change information. The notation should adhere to the format: GENE,NM_123456:c.-123N>N (no spaces are allowed). Once added, this information becomes available for the report.

• Exon information: Exon number and total number of exons for the selected transcript

• External resources: Quick links to UCSC Genome Browser, GeneCards, PubMed, WikiGenes, Genomenon

• dbSNP ID and link (SNV/Indel only)

• CNV-specific data:

  • SV Type (e.g., DEL, DUP)

  • SV Length

  • DECIPHER link

  • ISCN notation

  • Cytoband location

• Variant chips:

  • Homology Region:

    In Emedgene, variants located in regions of high sequence similarity are labeled with the Homology Region tag, reflecting potential mapping ambiguity due to paralogous sequences. This tag is automatically applied when High Sensitivity Mode is enabled (default), helping users identify variants that may require orthogonal validation. Users can also create preset filters using this tag to streamline variant triage in complex genomic regions.

  • Potential Mosaic:

    Emedgene now applies the Potential Mosaic tag to variants identified by its improved mosaic calling model, which delivers 4× fewer false positives than DRAGEN 4.2 High Sensitivity Mode and outperforms other mosaic callers in both speed and accuracy. Mosaic detection is enabled by default when running DRAGEN through Emedgene, with an allele frequency threshold of 0.2, and users can streamline interpretation by creating preset filters using this tag.

  • Ambiguous Calling:

    Emedgene now tags variants from segmental duplication regions with Ambiguous Calling, reflecting their uncertain placement due to poor mappability across ~5% of the genome. These variants are detected using the MRJD caller, which performs haplotype-based de novo small variant calling across paralogous regions, and are linked via a unique identifier (JIDS) in the new Connected Variants component. When DRAGEN is run through Emedgene, MRJD High Sensitivity Mode is enabled by default, and users can filter these variants using the Calling Methodology filter.

2. In silico Predictions

In silico Predictions highlight composite scores for Missense Prediction, Conservation, and Splicing Prediction. These are algorithmic assessments of variant effects based on known biological features, protein structures, evolutionary conservation, or machine learning models trained on large-scale data. The scores are calculated by proprietary algorithms that integrate outputs from individual in silico variant pathogenicity predictors.

These scores support ACMG classification, particularly PP3 (pathogenic evidence) and BP4/BP7 (benign evidence), and are especially useful when experimental data is lacking.

Each prediction type is grouped into three key categories: Missense Prediction, Conservation, Splicing Prediction.

Missense Prediction

Predicts whether a missense variant disrupts protein function.

Tools include:

• PolyPhen-2 (HDIV & HVAR) – Predicts the possible impact of an amino acid substitution on the structure and function of a protein (Adzhubei et al., 2010).

• SIFT – Evaluates whether an amino acid change affects protein function based on sequence homology (Ng & Henikoff, 2003).

• MutationTaster – Integrates conservation, splice site changes, and protein features to assess deleteriousness (Schwarz et al., 2014).

• LRT – Likelihood Ratio Test; compares selective constraint across species (Chun & Fay, 2009).

• DANN – Deep learning model trained on pathogenic and benign variants (Quang et al., 2015).

• REVEL – Ensemble predictor combines scores from 13 tools to assess missense variant pathogenicity with high sensitivity and specificity (Ioannidis et al., 2016).

• PrimateAI-3D – Developed by Illumina, enhances variant interpretation in Emedgene by leveraging deep learning model trained on 233 primate genomes. Its pathogenicity scores improve clinical prioritization of missense variants with structural and evolutionary precision (Sundaram et al., 2018).

• APOGEE – Predicts pathogenicity of variants using supervised learning and protein structure features (Diroma et al., 2021).

These tools help in prioritizing variants for review when missense changes are present. These also support ACMG PP3/ BP4 tagging based on REVEL thresholds. Combining the output of these tools with phenotype match to strengthen interpretation.

Conservation

Assesses how conserved a nucleotide or amino acid position is across different species, assuming that critical functional regions evolve slowly.

Tools include:

• SiPhy 29 Mammals – Detects regions under selective constraint (Garber et al., 2009).

• GERP RS – Measures evolutionary constraint by identifying regions where variation is suppressed (Davydov et al., 2010).

• phastCons 100 Vertebrates – Indicates conservation level using a phylogenetic hidden Markov model (Siepel et al., 2005).

These help in evaluating whether a variant lies in a functionally important region. They also support PP3 when conservation is high across multiple tools. Combined with missense scores accounts for stronger evidence.

Splicing Prediction

Determines whether a variant disrupts normal RNA splicing, potentially altering gene expression. Crucial for intronic and synonymous variants.

Tools include:

• dbscSNV-AdaBoost and RandomForest – Machine learning models for splice site impact (Jian et al., 2014).

• SpliceAI – Deep neural network trained on splicing data. Provides directional delta scores (Jaganathan et al., 2019):

  • DS_AG (Donor Gain)

  • DS_AL (Acceptor Loss)

  • DS_DG (Donor Loss)

  • DS_DL (Acceptor Gain)

• SpliceAI-10K – Evaluates effects from the variant, detecting large-scale effects like pseudoexonization, partial intron retention, and exon skipping. Scores are provided for donor and acceptor gain/loss, and high values may indicate cryptic splice site activation. These insights are used to strengthen ACMG tagging (e.g., PVS1 or PP3) and refine transcript-level interpretation, particularly for deep intronic variants or cases with suspected splicing disruption (Canson et al., 2023).

Outcomes of these tools help in flagging intronic or synonymous variants with high splicing impact. It supports PP3 or BP7 tagging based on SpliceAI thresholds. Helps prioritize variants for transcript-level review or RNA validation.

Currently available in silico predictions per variant type:

	SNV	Indel	mtDNA (SNV/indel)
Missense Prediction	Polyphen2 HDIV Polyphen2 HVAR SIFT MutationTaster LRT DANN REVEL PrimateAI-3D (34.0+)		APOGEE MitoTIP
Conservation	SiPhy 29 Mammals GERP RS phastCons 100 vertebrate	GERP RS
Splicing Prediction	dbscSNV-RF dbscSNV-Ada SpliceAI DS AG SpliceAI DS AL SpliceAI DS DG SpliceAI DS DL Predicted effect (SpliceAI-10k, 38.0+)

Note: Variants of types CNV, SV and STR are not annotated with in silico predictions.

3. Gene Metrics

Provides intolerance metrics from ExAC and gnomAD that relate to gene constraint and variant burden.

p(LoF intolerant)

pLI = p(LoF intolerant) is a probability of being loss-of-function intolerant to heterozygous and homozygous LoF variants.

Scale:

• 🔴 pLI ≥ 0.9: extremely LoF intolerant,

• 🟠 pLI > 0.1 & < 0.9: intermediate value,

• 🟢 pLI ≤ 0.1: LoF tolerant.

Z missense

The Z missense score indicates intolerance to missense variants based on the deviation of observed missense variants versus the expected number.

Scale:

• 🔴 Z missense ≥ 3: missense intolerant,

• 🟠 Z missense > 2.5 & < 3: intermediate value,

• 🟢 Z missense ≤ 2.5: missense tolerant.

p(REC)

p(REC) is a probability of being intolerant to homozygous, but not heterozygous LoF.

Scale:

• 🔴 p(REC) ≥ 0.8: Hom LoF intolerant,

• 🟠 p(REC) > 0.2 & < 0.8: intermediate value,

• 🟢 p(REC) ≤ 0.2: Hom LoF tolerant.

RVIS ratio

RVIS = Residual Variation Intolerance Score is indicative of a gene's intolerance to functional variation based on comparing the overall number of observed variants in a gene to the observed common functional variants.

Scale:

• 🔴 RVIS ≤ 30: functional variation intolerant,

• 🟠 RVIS > 30 & < 50: intermediate value,

• 🟢 RVIS ≥ 50: functional variation tolerant.

pLoF o/e

O/E Score is the ratio of the observed/expected number of LoF variants. It is a continuous measure of gene tolerance to LoF variation that incorporates a 90% confidence interval. The closer the O/E is to zero, the more likely the gene is LoF-constrained. If a hard threshold is needed for the interpretation of Mendelian disease cases, use the upper bound of the O/E confidence interval < 0.35.

Note: Gene Metrics are not available for CNVs or mtDNA variants.

4. Gene's related diseases

as reported in OMIM, ORPHANET, CGD, ClinVar, and academic papers included in the Emedgene knowledge graph. Each of the entries is provided with an inheritance mode icon and a link to the source.

Emedgene allows the customization of the disease name associated with a gene when editing from the Variant Page. The Customize Disease Name feature allows you to:

1. Edit the disease selected for a gene on the Variant Page.

2. Create a custom disease name for reporting purposes when none of the predefined options are appropriate.

This helps ensure that the disease name in your report is accurate, relevant, and tailored to the case you are working on.

Selecting a disease from the associated gene-disease list:

When you open the Gene Related Disease card on the Variant Page:

• Emedgene displays all known diseases associated with the gene, along with:

  • Inheritance mode (e.g., AD, AR, XLR).

  • Direct links to OMIM, CGD, and Orphanet database entries.

• If your variant is located in multiple genes, the list shown will contain existing associations per gene. You can select the one disease that applies from the association list.

• For tagged variants, you can choose a gene-related disease from the existing association list.

Default behavior:

• If you make no selection, the default associated disease will be used for reporting.

Creating a Custom Disease name:

If none of the associated diseases are appropriate for reporting or export:

1. Click the plus (+) button within the Gene Related Disease component.

2. Enter your custom disease name.

3. This name will:

   • Be stored in the case only.

   • Appear in the Summary section and in the report.

   • Not be stored in Curate (emedgene Curate currently does not support custom gene–disease connections or custom disease names).

Modifying the disease selection:

When you change the selected gene-related disease:

• The evidence graph will automatically update to reflect the new disease association.

• The system will prompt you to save or cancel your changes.

• The Gene Related Disease card will also show a pie chart comparing:

  • Disease-matched phenotypes vs. Disease phenotypes for the selected association.

• The change will be recorded in the Activity Log.

Impact on analysis and reporting:

• If you re-analyze the case:

  • AI Shortlist tags and candidate data from the previous analysis will be removed.

  • All user tags and user candidate data will be preserved.

• You can export the variant to Curate:

  • If you selected a disease from the predefined list → Curate will save that connection.

  • If you selected a custom disease → No disease will be saved in Curate for that variant.

Note: Only users with the edit evidence role can:

• Select a gene-related disease from the list.

• Create and assign a custom disease name.

Warning: Always double-check your disease selection before finalizing the report, as it directly affects evidence graphs, phenotype matching, and Curate data consistency.

Lists gene-disease connections from curated databases:

• OMIM, Orphanet, CGD, ClinVar

• Academic literature in Emedgene knowledge graph

Each entry includes:

• Disease name

• Inheritance mode icon (e.g., AD, AR, XLR)

• Source link

Use this card to verify known gene-disease relationships and potential implications of a variant in clinical context. The user can either select the disease from the available list of gene-disaese relationships or choose to add a custom disease.

5. Clinical significance

highlights previous pathogenicity classifications of the variant under review:

• Manually Classified indicates if the variant has been previously classified in any of the organization's cases by any user.

• Networks Classified indicates if the variant has been previously classified by the partnering organizations in your network.

• Curate indicates if the variant has been previously classified in your Curate variant database.

• ClinVar provides a list of ClinVar submissions for the selected variant.

• ClinGen Regions (only for CNVs) indicates whether a variant overlaps the established dosage-sensitive region defined by ClinGen.

• Custom database shows a variant class from the variant database(s) curated by your organization. We can easily implement an organization's curated database of classified SNV or CNV variants to facilitate the case review.

• MITOMAP shows a variant's status in MITOMAP. By clicking on the MITOMAP interactive link, you will be taken to MITOMAP: Reported Mitochondrial DNA Base Substitution Diseases: Coding and Control Region Point Mutations.

Caution: Please be aware that there might be instances where the Variant page > Clinical significance > Networks classified section appears erroneously empty. However, you can still rely on the Variant page > Related cases, which will continue to display relevant information as intended. Please utilize the Variant page > Related cases section while the fix is being implemented.