Predicting amyloid aggregation rates of proteins using multivariate analysis
(English)Manuscript (preprint) (Other academic)
Several diseases have been linked to the presence of extracellular protein deposits of β-rich aggregates, known as amyloid fibrils. The formation of these fibrils and their precursors has been identified as key players in the development of these diseases. It is therefore desirable to gain a deeper understanding of the mechanism of amyloid aggregation.In this study we have used multivariate analysis to elucidate the most important physicochemical and structural factors of amino acids that are important for the amyloid aggregation. We used a combination of principal component analysis, orthogonal partial least squares and auto- and cross-covariance to investigate a database consisting of amyloid aggregation rate measurements of 77 AcP mutants.Our results show that changes in hydrophobic patterns, charge and β-sheet propensity is common for mutants with the largest changes in amyloid propensity. In addition, we can also, with reasonable accuracy, predict the amyloid aggregation rate of a test set of AcP mutants that were not used to create the initial aggregation model. Thus, the multivariate approach used in this study is shown to be powerful tools to extract important factors of protein amyloid aggregation that are hidden in the growing pool of available experimental data of amyloid aggregation.
Multivariate analyis, amyloid aggregation, OPLS, AcP
IdentifiersURN: urn:nbn:se:umu:diva-46591OAI: oai:DiVA.org:umu-46591DiVA: diva2:439273