TRACE: A Framework for Integrating Transcript Relevance Into ACMG/AMP Variant Interpretation.

Background: Accurate clinical variant interpretation depends on the transcript used for annotation and consequence assessment. Transcript-aware reasoning is also incorporated into existing ClinGen guidance for loss-of-function, splicing, functional and computational evidence and into gene- and disease-specific specification. However, limited guidance exists when unresolved transcript context warrants escalation beyond routine annotation, how heterogeneous transcript-relevance evidence should be integrated or how the resulting determination should be documented. Methods: We propose TRACE (Transcript Relevance Assessment for Clinical Evaluation), a four-tiered gated framework. Applicable gene- or disease-specific Variant Curation Expert Panel (VCEP) specifications take precedence where they resolve the transcript question. Otherwise, TRACE begins with standard transcript annotation, escalates to focused transcript-relevance assessment only when transcript context could materially alter molecular consequence or criterion applicability and reserves targeted developmental, RNA, protein or functional evidence for unresolved cases. Results: TRACE separates transcript-context assessment from criterion-specific evidence assignment. Transcript context may establish whether the biological prerequisite for PVS1, PM1, PM4, PS3/BS3 or PP3/BP4 assessment is satisfied; established ClinGen SVI or gene-specific guidance then determines whether the criterion is applied and at what strength. Representative mechanisms include exon utilisation in Conclusions: TRACE is an escalation and documentation framework, not a parallel evidence-weighting system. Its primary aim is to make clinically material transcript determinations explicit, reproducible and auditable while preserving established SVI/VCEP rules for evidence application and strength. Whether TRACE improves classification accuracy, interlaboratory concordance or diagnostic yield requires empirical validation.