RENOVO-NF1 accurately predicts NF1 missense variant pathogenicity

Area of Science:

• Genetics
• Computational Biology
• Medical Diagnostics

Background:

• Identifying pathogenic variants in the NF1 gene is crucial for diagnosing neurofibromatosis, but is often challenging due to factors like allelic heterogeneity and lack of functional assays.
• Existing computational tools are not yet established for NF1 variant interpretation, highlighting a need for specialized solutions.

Purpose of the Study:

• To optimize and validate RENOVO, a random forest-based predictor, for accurate interpretation of NF1 variants.
• To assess RENOVO's performance in classifying variants with insufficient information for immediate diagnostic use, mimicking real-world clinical scenarios.

Main Methods:

• RENOVO was developed using "database archaeology," analyzing historical ClinVar data to identify "stable" variants (consistent classification) for training and "unstable" variants (reclassified from VUS) for testing.
• Performance was validated on two independent sets: ClinVar reclassifications of initially unknown significance (VUS) variants and de novo variants from a clinical cohort classified per ACMG criteria.

Main Results:

• RENOVO achieved high accuracy across datasets: 98.6% on training, 96.5% on testing, 82% on validation set 1 (96.2% for missense variants), and 93.7% on validation set 2.
• The tool demonstrated consistent performance, indicating its reliability in interpreting NF1 variants with limited initial data.

Conclusions:

• RENOVO-NF1 accurately interprets NF1 variants, particularly those lacking sufficient information for standard ACMG classification at the time of detection.
• This computational tool shows significant potential to address diagnostic challenges in neurofibromatosis type 1.