Question 1

How does AI-driven variant classification differ from manual ACMG interpretation?

Accepted Answer

Manual ACMG interpretation requires clinical teams to gather population frequencies, functional prediction scores, splice impact evidence, literature references, and internal laboratory history for each variant. The AI-driven variant classification platform automates evidence aggregation from sources such as gnomAD, ClinVar, REVEL, CADD, and SpliceAI, presenting interpreters with a suggested ACMG classification, confidence score, supporting rationale, and auditable evidence trail. This approach helps clinical teams reduce interpretation time while maintaining human review and final approval.

Question 2

What data sources does the platform aggregate for variant interpretation?

Accepted Answer

The platform aggregates evidence from population databases including gnomAD, ClinVar, and 1000 Genomes; functional prediction tools such as REVEL, CADD, SIFT, PolyPhen-2, and AlphaMissense; splice prediction tools including SpliceAI and MaxEntScan; oncology knowledge bases such as OncoKB, CIViC, and COSMIC; and internal laboratory classification histories. It can also incorporate phenotype data through Human Phenotype Ontology (HPO) terms to support phenotype-driven variant prioritization.