A Novel Method for Lysine Acetylation Sites Prediction

As one of the most important and complex post translational modifications, acetylation modulates a number of cellular processes. Experimental identification of acetylation sites is still tricky and time-consuming. Thus, prediction of acetylation sites through computational models will facilitate functional annotations of the human proteome. In this work, by selecting amino acid (AA) and dipeptide (DIP) compositions as parameters, an excellent algorithm, namely, Increment of Diversity with Quadratic Discriminant analysis (IDQD) was proposed to predict internal lysine acetylation sites. An accuracy of 94.38% was obtained in the jackknife test, superior to that of the support vector machine. Furthermore, we found that the accuracy was independent of window length. These results suggest that our IDQD model will play important roles in the realm of acetylation sites identification.