Abstract

A web-based tool namely N-Ace can predict protein acetylation sites depend amino acid sequence and other structural characteristics, such as accessible surface area, absolute entropy, non-bonded energy, size, amino acid composition, steric parameter, hydrophobicity, volume, mean polarity, electric charge, heat capacity and isoelectric point which is surrounding the modification site. Based on the concepts of Support Vector Machine (LIBSVM), computational models are learned from the amino acids sequence, structural characteristics, and physicochemical properties. Moreover, we construct independent test dateset, which remove the same datasets from UniProtKB/ Swiss-Prot v55 and dbPTM, for independent test, and we show that the acetylation prediction accuracies on alanine, lysine and serine are 92%, 90% and 81%, respectively. Therefore, this work proves that N-Ace can perform better than other prediction tools.

System Flow

Data Statistics