Prediction of protein N-formylation using the composition of k-spaced amino acid pairs

As one of important protein post-translational modifications, N-formylation has been reported to be involved in various biological processes. The accurate identification of N-formylation sites is crucial for understanding the underlying mechanisms of N-formylation. Since the traditional experimental methods are generally labor-intensive and expensive, it is important to develop computational methods to predict N-formylation sites. In this paper, a predictor named NformPred is proposed to improve the prediction of N-formylation sites by using composition of k-spaced amino acid pairs encoding scheme and support vector machine algorithm. As illustrated by 10-fold cross-validation, NformPred achieves a promising performance with a Sensitivity of 86.00%, a Specificity of 96.25%, an Accuracy of 94.48% and a Matthew's correlation coefficient of 0.8099, which are much better than those of current computational method. Feature analysis shows that some k-spaced amino acid pairs such as 'IxxL', 'LV' and 'IxxxI' play the most important roles in the prediction of N-formylation sites. These predictive and analytical results suggest that NformPred might facilitate the identification of protein N-formylation. A free online service for NformPred is accessible at http://123.206.31.171/NformPred/.