ACHP: A Web Server for predicting anti-cancer peptide and
anti-hypertensive peptide
Abstract
Peptide drugs are generally compounds with less than 100 amino acids
connected by peptide bonds and having drug effects. Because of their
unique advantages such as significant activity, strong specificity, and
low toxicity, they are widely applied in the treatment of various
diseases. The design and development of new peptide drugs have broad
prospects, and determining the molecular characteristics of
disease-related peptide drugs is the key to drug design. This research
takes anti-cancer peptides and anti-hypertensive peptides as the
research objects, and we propose a novel method of describing peptide
drugs, making use of the topological attribute values in an amino acid
interaction network to represent the characteristics of peptides. In
addition, peptide drugs are described from different perspectives by
combining the information of the primary, secondary and tertiary
structures. Three algorithms including support vector machine (SVM),
K-nearest neighbor (KNN) and random forest (RF) are utilized to train
the model. Then the support vector machine based on recursive feature
elimination method (SVM-RFE) removes redundant features and identifies
the key characteristics of different types of drugs. The added network
features can more comprehensively describe peptide drugs, providing a
theoretical basis for the analysis and design of new peptide drugs. The
web sever of ACHP is freely available at http://118.178.58.31:9801/.