< Back to previous page


Computational prediction and interpretation of the molecular effect and disease phenotype of missense variants in the human exome

Book - Dissertation

Despite the advent of new high-throughput experimental methods and the incredible amount of data they produce, it remains challenging to understand the link between inherited diseases, their genetic causes, and the impact of variants on protein structure, stability, and function. While many bioinformatics tools predict which variants in the human exome are likely to cause diseases, few tools predict the molecular causes of these phenotypes. Nonetheless, this information is necessary for the rational development of drugs that alleviate the symptoms of genetic diseases and for the development of personalized treatments.

This thesis contributes to filling this gap with the development of two computational tools, SNPMuSiC and FuncMuSiC, that use protein 3-dimensional structure information to predict the deleterious effect of missense variants in the human exome. Both classify variants as disease-causing or neutral on the basis of biophysical characteristics: SNPMuSiC is based on changes in protein stability caused by the variant and Func-MuSiC, on the proximity between the variant residue and annotated functional sites. By construction, these two algorithms help improve the interpretation of the disease phenotype at the molecular level.

We also analyzed the case of sphingomyelin phosphodiesterase (SMPD1), a key enzyme in the human sphingolipid metabolism which plays a central role in eukaryoticplasmic membranes turnover. Several mutations in this enzyme are known to cause the Niemann-Pick disease, a syndrome characterized by different phenotypes ranging from severe symptoms impacting the central nervous system (type A) to milder symptoms impacting viscera only (type B). By combining evolutionary, structural, and contextual information, we developed the SMPD1-ZooM predictor, which predicts whether anamino-acid variant in SMPD1 is neutral or associated with type A or type B disease.This predictor thus offers information about both the severity of the disease and the molecular phenotype.

The results obtained in this thesis are available online on several webservers that we developed. They are an important step towards the prediction and interpretation of the molecular effect of missense variants and towards the understanding of their link with inherited diseases. Moreover, these results offer interesting perspectives for disease diagnostic and the development of targeted therapeutic approaches.
Number of pages: 172
Publication year:2022