The prediction of palmitoylation site locations using a multiple feature extraction method
As an extremely important and ubiquitous post-translational lipid modification, palmitoylation plays a significant role in a variety of biological and physiological processes. Unlike other lipid modifications, protein palmitoylation and depalmitoylation are highly dynamic and can regulate both protein function and localization. The dynamic nature of palmitoylation is poorly understood because of the limitations in current assay methods. The in vivo or in vitro experimental identification of palmitoylation sites is both time consuming and expensive. Due to the large volume of protein sequences generated in the post-genomic era, it is extraordinarily important in both basic research and drug discovery to rapidly identify the attributes of a new protein's palmitoylation sites. In this work, a new computational method, WAP-Palm, combining multiple feature extraction, has been developed to predict the palmitoylation sites of proteins. The performance of the WAP-Palm model is measured herein and was found to have a sensitivity of 81.53%, a specificity of 90.45%, an accuracy of 85.99% and a Matthews correlation coefficient of 72.26% in 10-fold cross-validation test. The results obtained from both the cross-validation and independent tests suggest that the WAP-Palm model might facilitate the identification and annotation of protein palmitoylation locations. The online service is available at http://bioinfo.ncu.edu.cn/WAP-Palm.aspx. A new computational method WAP-Palm, combining multiple feature extraction, is designed to identify protein palmitoylation site. The multiple feature descriptors are composed of weight amino acid composition (WAAC), auto-correlation functions (ACF) and position specific scoring matrix profiles (PSSM), respectively. WAAC is utilized to extract amino acids sequence position information surrounding palmitoylation sites. ACF is applied to encode the physicochemical properties and the correlation of amino acid residues. PSSM is used to represent evolutionary information around palmitoylation sites. The WAAC, ACF and PSSM features all contributed to the palmitoylation prediction. The WAP-Palm improved the quality of identifying palmitoylation sites and might facilitate the annotation of protein palmitoylation. âº Our predictor achieves a Matthews correlation coefficient of 72.26%. âº The residues around palmitoylation sites have better correlation and dependence. âº Evolution information acts an irreplaceable role for palmitoylation prediction. âº Positively charged residues are enriched at the downstream of palmitoylation sites.