| Zusammenfassung |
We study how mutations affect proteins, focusing on two closely related questions: 1) How likely is it that a given protein variant has a clinically relevant effect; and 2) What is the protein-level molecular mechanism by which a variant causes disease? To do this, we employ three highly complementary strategies. Computational variant effect predictors, driven primarily by evolutionary information, are very good at identifying pathogenic mutations in certain genes, but tell us nothing about why they are damaging. In contrast, using structural bioinformatics to investigate the 3D protein context of mutations can provide great insight into the molecular mechanisms underlying disease mutations, but has historically been less useful for identifying deleterious mutations. Finally, deep mutational scanning, which allows direct high-throughput measurement of variant effects, is proving tremendously valuable for the identification disease mutations, and can also clarify molecular mechanisms given the correct experimental design. Our focus is on developing optimal strategies for utilising all three approaches to most efficiently and effectively identify damaging protein variants and elucidate their molecular mechanisms, ultimately leading to improved diagnosis and treatment of human genetic disease. |
