AnnotSV โ Between Annotation and Clinical Classification
Why is this important?
Structural variants (SVs) and copy-number variants (CNVs) can encompass multiple genes and regulatory regions within a single coordinate interval. When breakpoints are imprecise or different callers employ disparate representations, manually cross-referencing all databases becomes impractical. AnnotSV annotates variant coordinates with gene, population, disease, and dosage-sensitivity information, organizing them into an interpretable table for review.
While annotation accelerates interpretation, it does not eliminate the uncertainty inherent in raw data. Events missed or breakpoints incorrectly called by the input caller cannot be corrected through rich annotation alone.
Scope Determined by Input
AnnotSV accepts the chromosome, start, end, and type of SVs/CNVs expressed in formats such as BED and VCF. Because differing genome builds may cause identical numerical coordinates to refer to entirely different genomic regions, the build must be specified accurately. The callerโs filters, confidence intervals, and sample genotypes must also be preserved.
SV representations vary among callers; a single complex event may be split into multiple records, or reciprocal events may be recorded in opposite directions. Without normalization and sample-level review, one row cannot be definitively equated to one event.
What information to include
The tool can integrate overlapping genes and transcripts, known pathogenic regions, population structural variants (SVs), phenotype and disease databases, and dosage evidence such as haploinsufficiency and triplosensitivity. Complete inclusion and partial overlap have distinct meanings; when only a single exon is affected, the transcript and reading frame must be evaluated separately.
The absence of an entry in a database may suggest rarity but does not prove pathogenicity. Conversely, even if there is overlap with a population database, it may not represent the same benign event if the size, breakpoints, phenotype, or zygosity differ.
Ranking and Clinical Classification
Automated ranking is a tool that prioritizes candidate variants for review. The ACMG/ClinGen CNV criteria structure the evaluation of gene content, dosage sensitivity, population evidence, inheritance, and phenotype; however, they incorporate professional judgment. AnnotSVโs class or score should not be directly copied as the final clinical classification.
Tools cannot fully ascertain patient phenotype, family segregation, and testing purpose. The de novo status, mosaicism, and the second allele of a recessive gene require separate review.
Small Example
Suppose a deletion overlaps with some exons of a dosage-sensitive gene. Although the annotation table may display the gene name and associated disease, it is necessary to verify whether the reading frame of the major transcript is disrupted, whether the breakpoint lies outside the exon, and whether other transcripts retain function. Furthermore, the certainty of breakpoints differs between low-resolution arrays and sequencing-based methods.
Even if the same coordinate is observed in a parent who appears phenotypically healthy, possibilities remain for reduced penetrance, variable expressivity, and unassessed parental phenotype.
Reproducibility
Record the AnnotSV version, as well as the annotation database release and download date. This distinguishes whether differences in results before and after updates stem from changes in knowledge or software. Preserve the original VCF, normalized input, command options, and outputs to enable auditing.
Common Misconceptions
- Annotation does not substitute for variant detection or orthogonal confirmation.
- Gene overlap is not equivalent to gene disruption.
- A rank of 1 does not confirm the causal variant.
- The automated class differs from the final classification by a clinical expert.
Interpretation Boundaries
The results depend on the input caller, genome build, breakpoint uncertainty, and database snapshot. Clinical interpretation requires a comprehensive review of phenotype, inheritance pattern, assay resolution, and current ACMG/ClinGen criteria.
Reading in Context
CNV confirmation methods are linked to qPCR CNV, assay-specific SV blind spots are linked to WGS, WES, and panel, and reference coordinate issues are linked to the Reference genome concept.