Prediction Of Post-translational Modification Cross-talk Within Proteins Using Residue-and Residue Pair-based Features

Post-translational modification(PTM)cross-talk plays a crucial role in regulating biological processes,including protein activity,cell signal transduction,gene expression,and protein-protein interactions.Studying this cross-talk is useful for further clarifying the regulation mechanism mediated by PTM.It is time-consuming and laborious to detect PTM cross-talk by experimental methods,and the development of calculation methods is expected to make up for the shortcomings of experimental techniques.Most of the existing computational studies rely on the correlation features between residues at the sequence level to develop prediction models,but neglected the structural information of cross-talk pairs and the characteristics of individual residues involved in cross-talk,which may restrict the improvement of the prediction accuracy.Therefore,developing novel algorithms is critical to overcome the current limitations.In this study,we propose a structure-based algorithm(PTM Cross-Talk predictor,PCTpred)to improve the accuracy of PTM cross-talk prediction.The algorithm first designed a series of residue pair-based features(such as co-evolution information,colocalization information,etc.)and residue-based features(such as deleterious score,Laplacian norm,topological index,etc.)at the protein sequence and structure level.Through comparative analysis,it is found that positive and negative samples have significant differences in residue pair-and residue-based features.Then,using the forward feature selection technique,we reserved 23 newly introduced descriptors and 3 traditional descriptors to establish a sequence-based predictor PCTseq and a structure-based predictor PCTstr,both of which were weighted to construct our final prediction model.According to sample-and protein-based evaluations,PCTpred yielded area under the curve values of 0.903 and 0.804,respectively.Even when removing the distance preference of samples or using the input of modelled structures,our prediction performance was maintained or moderately reduced.When testing different types of PTM cross-talk subsets and the data from the co-modified peptides collected in the literature,PCTpred still achieved good prediction results,thus demonstrating the strong generalization ability of our algorithm.Compared with the state-of-the-art methods,PCTpred can achieve higher prediction accuracy in various evaluations.The source code and dataset are freely available at https://github.com/Liulab-HZAU/PCTpred.