Missense variants pathogenicity annotation from homologous proteins

Area of Science:

• Genomics
• Human Genetics
• Bioinformatics

Background:

• High-throughput sequencing identifies millions of human genome single nucleotide variants (SNVs), with few linked to disease.
• Interpreting missense variants, which alter protein sequences, is difficult due to limited clinical and experimental data.
• Existing prediction tools struggle with variant pathogenicity due to reliance on conservation and structural information.

Purpose of the Study:

• To investigate the pathogenicity of homologous missense variants in proteins implicated in autosomal dominant diseases.
• To develop a robust method for predicting missense variant pathogenicity.
• To improve genotype-phenotype correlations and rare genetic disorder diagnosis.

Main Methods:

• Analysis of 2976 pathogenic and 17,555 non-pathogenic homologous variants.
• Development of the HomolVar web server for computational prediction of variant pathogenesis using homologous variant annotations.
• Evaluation of 27 common mutation predictor methods.

Main Results:

• Pathogenicity prediction accuracy reached 95% within families and 98% for closer homologs.
• Homologous variant analysis demonstrated a biological feature not fully captured by existing mutation predictors.
• The HomolVar web server was developed and made freely available at https://rarevariants.org/HomolVar.

Conclusions:

• Homologous missense variant analysis provides a robust method for predicting variant pathogenicity.
• The HomolVar tool enhances the prediction of unannotated variant effects.
• This approach aids in diagnosing rare genetic disorders and understanding genotype-phenotype relationships.