Choosing a Variant Annotation Platform

A raw VCF file tells you where a variant is and what it changed - it says nothing about whether that change matters. Annotation cross-references each variant against external databases to add that context, and the breadth and quality of those sources is what separates a useful annotation platform from a thin one.

• Population frequency: how common is this variant in the general population or specific ancestries? Rare variants are more likely to be disease-relevant.
• Clinical significance: has this exact variant been reported and classified before?
• Functional impact prediction: computational scores estimating whether a missense change is likely damaging.
• Gene and transcript context: which gene, which transcript, which protein domain is affected?
• Splice site and regulatory impact: does the variant affect splicing rather than the coding sequence directly?

• How many independent source categories does annotation draw from - population, clinical, functional, gene-level - not just a count of databases?
• Can annotation be scoped to a custom gene panel for a specific indication, or only run genome-wide?
• Is the reference genome build (GRCh37 vs GRCh38) explicit and consistent across every annotation source used?
• How is batch throughput handled for cohort-scale studies rather than single samples?