Identifying Ion Ligand Binding Sites With The Energy,Physicochemical And Structural Features

In many important life activities,the performance of protein functions depends on the interaction between proteins and ligands.As an important protein binding ligand,the recognition of binding sites is of great significance for the study of protein functions.The theoretical prediction method shows good prospects for identifying the ion ligand binding sites and is also the development direction of theoretical biophysics.In this paper,based on protein sequence information,ion ligand binding sites are identified by fusing energy,physicochemical and structural features.The main work is as follows:?1?The datasets of four acid radical ion ligand binding residues including NO₂^-,CO₃^2-,SO₄^2-and PO₄^3-were constructed,and the optimal window lengths of four acid radical ion ligands were determined as 11,13,11 and 9 by the sliding window method.the four acid radical ion ligand and ten metal ion ligand binding residues were taken as the research objects.?2?Based on the protein sequence information,according to the biological background,the amino acid,hydrophobicity,charge,secondary structure and relative solvent accessibility area were selected as the basic characteristic parameters.Then we extracted their component information,2L-dimensional?L is the optimal window length?position conservation information and refinement characteristic parameters,and used them as predicted feature parameters,respectively,through the sequence minimum optimization?SMO?algorithm identified the binding sites of 14 ion ligand with 5-fold cross-validation.It was found that the recognition results under the refinement characteristic parameters were the best.?3?Since the complex is in a stable state when the protein is combined with ion ligand,according to the principle of lower energy and the complex more stable,the Laplace energy of amino acids was introduced as a new basic characteristic parameter,and it was divided into 5 categories according to its value,then we extracted its component information and 2L-dimensional position conservation information as predicted feature parameters.Combined the hydrophilic and hydrophobic from the 6classifications into 4 classifications,and proposed the method of information entropy to extract its information.The component information and 2L information of amino acid,Laplace energy,charge,secondary structure,and relative solvent accessibility area were integrated into the support vector machine?SVM?algorithm with fusing the information entropy characteristics of hydrophilic and hydrophobic,14 ion ligand binding sites were identified by 5-fold cross-validation and obtained good prediction results.In order to verify the stability of the prediction model,an independent test was carried out and the results were better than the predecessors.In addition,based on fewer feature parameters,the results were better than the previous independent test results.?4?An online server for predicting and identifying the binding sites of ion ligands was established,which is free and open to the public.