Multimodal Genotype-Phenotype Analysis in SMARCB1-Associated Developmental Disorders.

Area of Science:

• Genetics
• Molecular Biology
• Developmental Biology

Background:

• SMARCB1 gene variants are linked to intellectual developmental disorders like Coffin-Siris Syndrome (CSS).
• The precise relationship between SMARCB1 variant location and specific phenotypes is not fully understood.
• Understanding genotype-phenotype correlations is crucial for improved diagnostics and patient care.

Purpose of the Study:

• To investigate the correlation between the location of SMARCB1 variants and associated phenotypic manifestations in individuals with CSS.
• To define the genotype-phenotype spectrum for SMARCB1-related disorders more comprehensively.

Main Methods:

• Analysis of 31 individuals with pathogenic SMARCB1 variants using multimodal approaches.
• Integration of clinical data, structural analysis (3D protein modeling), and machine learning (GestaltMatcher, XGBoost).
• Prediction of variant effects and phenotype-driven genotype classification.

Main Results:

• SMARCB1 variants were found to cluster in N-terminal and C-terminal regions.
• C-terminal variants correlated with more severe speech delay, microcephaly, and cleft palate, showing stronger facial similarity.
• Machine learning models achieved high accuracy (96.7%) in classifying variant locations from phenotypes, with clinical features being more predictive for N-terminal variants.

Conclusions:

• Distinct phenotypic differences exist between individuals with N-terminal versus C-terminal SMARCB1 variants.
• Multimodal assessment integrating phenotyping and machine learning enhances understanding of genotype-phenotype associations.
• This approach offers potential improvements for diagnostic strategies in SMARCB1-related disorders.