# Recognition of a Critical Functional Domain and Improved PHOX2B Missense Variant Interpretation by Utilization of In Silico Prediction Tools

## TL;DR

This study improves the interpretation of PHOX2B gene variants linked to a breathing disorder by using computational tools and identifying a critical region in the gene.

## Contribution

The study introduces a calibrated framework for PHOX2B missense variant classification using in silico tools and identifies the homeodomain as a critical functional region.

## Key findings

- BayesDel and REVEL are the strongest predictive tools for PHOX2B variant classification.
- Pathogenic variants cluster in the homeodomain, supporting its role as a critical functional domain.
- Using calibrated thresholds reduces the number of variants classified as uncertain significance.

## Abstract

Pathogenic heterozygous variants in PHOX2B are associated with congenital central hypoventilation syndrome (CCHS), which is characterized by autonomic nervous system dysregulation severely affecting respiratory control. The interpretation of PHOX2B missense variation is challenging due to their rarity and the lack of available functional evidence. Consequently, most PHOX2B missense variants are classified as variants of uncertain significance (VUSs), complicating the timely diagnosis and clinical management of the condition. To generate an improved model for assessments of PHOX2B missense variants, a methodology was derived to evaluate all PHOX2B missense variants in the literature and public/private databases according to a consensus classification framework and assigned pathogenicity classifications. Pathogenicity prediction scores from the in silico prediction tools CADD, REVEL, BayesDel, and AlphaMissense were obtained for all variants. A weighted logistic regression in a multiple imputation framework was performed to assess the strength of evidence supporting application of ACMG/AMP guidelines′ PP3/BP4 criteria. CADD, REVEL, and BayesDel meet the predictive strengths for PP3/BP4 recommended by the Clinical Genome Resource (ClinGen). Based on their areas under the curve and low proportions of variants with indeterminate pathogenicity predictions, BayesDel and REVEL were the strongest predictive tools and should be utilized for routine PHOX2B missense variant assessment with this study′s calculated score thresholds for PP3/BP4 strength levels. Furthermore, the positional distribution of pathogenic and benign variants was analyzed to assess potential hotspots or critical functional domains in PHOX2B, and pathogenic variants were found to cluster in the homeodomain. The enrichment of pathogenic variation was substantiated by the prediction tools, supporting the use of the PM1 criterion for variants in the homeodomain. This calibration of existing computational prediction tools for PHOX2B missense variant classification and recognition of the homeodomain variants will enable fewer VUS classifications in favor of conclusive results, aiding in these individuals′ care.