Using Multi‐Level Correlation Information to Identify Amyloidogenic Peptides

Abstract

AbstractIn recent years, more and more researchers draw their attention on exploring the characteristics of peptides. This study focused on identifying amyloidogenic peptides, since prior works reported that amyloidogenic peptides may keep a close relationship with neurodegenerative disease. To investigate this issue, we collected peptide sequences from Pep424 dataset. Fifty classes of physicochemical (PC) properties were employed to encode samples. Then, Pearson's correlation coefficient (PCC) was used to capture the correlation information among distinct PC properties. After that, we adopted the least absolute shrinkage and selection operator (LASSO) algorithm to select these most discriminative features. Next, these selected features were fed into support vector machine (SVM) for identifying amyloidogenic peptides. An accuracy of 89.62 % was obtained in jackknife test. A significant improvement was achieved by the proposed method, as compared with the state‐of‐the‐art predictors. It means that our method is effective and competitive in exploring amyloidogenic peptides field. We provide the dataset and Matlab codes for academic use, which can be available at https://figshare.com/articles/online resource/iAMY/17371877.