Bioinformatics Analysis Of Protein Hydroxylation

Hydroxylation is an important and reversible protein post-translational modification activity.It is critical for the stabilization of the triple helix conformation.The pattern of these modifications influences many biological functions,including fibrillogenesis,crosslinking and matrix mineralization.Hydroxyproline and hydroxylysine have been found to be implicated in metabolic disorder,connective tissue disorders,lung cancer and stomach cancer.First of all,identification of hydroxylation sites is a basic step to understand their molecular mechanism in biological systems.Because experimental identification of hydroxylation is time-consuming and expensive,bioinformatics tools with high accuracy represent desirable alternatives for large-scale rapid identification of protein hydroxylation sites.With the rapid development of bioinformatics technology,the method combined with computer technology and mathematics principle to predict hydroxylation sites has become a research hot spot.In this paper,we establish a model by using bioinformatics methods,mathematical principle based on a large amount of data and high performance computing platform.Proline and lysine are two main amino acid residues to be hydroxylated in proteins.In view of this,the protein sequences containing experimentally verified protein hydroxylation sites were collected from the UniProt/Swiss-Prot database.Total 265 candidate proteins containing hydroxylated pralines and 34 candidate proteins containing hydroxylated lysines were collected,respectively.Then,we developed a supporter vector machine-based tool,OH-PRED,for the prediction protein hydroxylation sites using the adapted normal distribution bi-profile Bayes feature extraction in combination with the physicochemical property indexes of the amino acids.These results demonstrate that OH-PRED outperforms previously published methods.All demonstrated the effectiveness of our model,with good guidance for future biological experiments.Given OH-PRED good prediction performance,we generalized it's methods to enhancer's recognition,build model EnhancerPred and EnhancerPred2.0,which also got good performance.