Prospective functional classification of all possible missense variants in PPARG

Area of Science:

• Genetics
• Molecular Biology
• Bioinformatics

Background:

• Clinical exome sequencing frequently identifies missense variants in disease-associated genes, posing diagnostic challenges due to lack of functional data.
• Mutations in PPARG are linked to Mendelian lipodystrophy and increased type 2 diabetes (T2D) risk, with many variants of unknown significance.

Purpose of the Study:

• To prospectively develop a method for functional characterization of all possible single-amino acid substitutions in the PPARG gene.
• To create a machine learning classifier trained on experimental data to interpret the pathogenicity of novel PPARG missense variants.

Main Methods:

• Saturation mutagenesis was employed using oligonucleotide synthesis to create a library of all 9,595 possible single-amino acid substitutions in PPARG.
• A pooled functional assay in human macrophages was developed to experimentally evaluate all protein variants.
• Supervised machine learning was used to train a variant classifier on the experimental data.

Main Results:

• The developed classifier successfully annotated 6 out of 55 newly identified missense variants as pathogenic.
• These pathogenic variants were subsequently validated through single-variant assays, confirming the classifier's accuracy.
• The study demonstrated the utility of saturation mutagenesis and experimental characterization for immediate diagnostic interpretation.

Conclusions:

• Saturation mutagenesis combined with functional assays and machine learning provides a powerful approach for interpreting missense variants in disease-related genes like PPARG.
• This methodology can significantly reduce diagnostic uncertainty for patients with genetic conditions.
• The findings support the integration of prospective experimental characterization into clinical genetic testing workflows.