I’m implementing CNV/SV annotation and seeing a common issue: OMIM gene to phenotype mappings explode for multi-gene events and swamp the output. Right now we map overlapped genes to OMIM phenotypes and score them against case HPO terms, but we still get long, low-specificity lists.

Inputs: CNV/SV coords, overlapped genes, case HPO terms, ClinGen dosage where available, and ClinVar/DECIPHER evidence when present.

What heuristics have worked for you to keep signal high?

• report-facing summary vs analyst table separation
• gene ranking (ClinGen dosage, OMIM morbid, HI/TS, known syndrome regions, phenotype match)
• phenotype display rules (HPO-matched only, tiered top-N, ClinGen-only)

Would appreciate any concrete strategies or examples from your pipelines.