Predicting phosphorylation sites using machine learning by integrating the sequence, structure, and functional information of proteins

Abstract

Background Post-translational modification (PTM) is a biological process that alters proteins and is therefore involved in the regulation of various cellular activities and pathogenesis. Protein phosphorylation is an essential process and one of the most-studied PTMs: it occurs when a phosphate group is added to serine (Ser, S), threonine (Thr, T), or tyrosine (Tyr, Y) residue. Dysregulation of protein phosphorylation can lead to various diseases—most commonly neurological disorders, Alzheimer’s disease, and Parkinson’s disease—thus necessitating the prediction of S/T/Y residues that can be phosphorylated in an uncharacterized amino acid sequence. Despite a surplus of sequencing data, current experimental methods of PTM prediction are time-consuming, costly, and error-prone, so a number of computational methods have been proposed to replace them. However, phosphorylation prediction remains limited, owing to substrate specificity, performance, and the diversity of its features. Methods In the present study we propose machine-learning-based predictors that use the physicochemical, sequence, structural, and functional information of proteins to classify S/T/Y phosphorylation sites. Rigorous feature selection, the minimum redundancy/maximum relevance approach, and the symmetrical uncertainty method were employed to extract the most informative features to train the models. Results The RF and SVM models generated using diverse feature types in the present study were highly accurate as is evident from good values for different statistical measures. Moreover, independent test sets and benchmark validations indicated that the proposed method clearly outperformed the existing methods, demonstrating its ability to accurately predict protein phosphorylation. Conclusions The results obtained in the present work indicate that the proposed computational methodology can be effectively used for predicting putative phosphorylation sites further facilitating discovery of various biological processes mechanisms.

Evaluating machine learning-powered classification algorithms which utilize variants in the GCKR gene to predict metabolic syndrome: Tehran Cardio-metabolic Genetics Study

“…Among classification machine learning methods that included SVM, DT and RF, RF had the best performance in classifying subjects on their MetS outcomes as indicated by the highest accuracy (0.743) as well as area under the receiver operating characteristic curve (AU-ROC) (0.804) and AUC-PR (0.776). This result is similar to the findings in the study by Szabo et al that applied the Random Forest algorithm for a similar task and calculated the accuracy of this method to be 71.4% [ 46 – 48 ]. Worachartcheewan et al also implemented a Random Forest model to predict MetS in the Bangkok population and identify the most influential predictors.…”

Shared and unique phosphoproteomics responses in skeletal muscle from exercise models and in hyperammonemic myotubes

“…To identify protein kinase-substrate relationships within the clusters ( Jamal et al., 2021 ), we performed substrate motif analysis for each temporal cluster of DEpP ( Figure S17 ). In the P. increase and decrease clusters , the recognition motifs for phosphorylation by PKA, PKC, casein kinase I, GSK3 and Ca2+/calmodulin-dependent protein kinase 2 (CAMK2) family members were noted.…”

Comparative Phosphoproteomics of Neuro-2a Cells under Insulin Resistance Reveals New Molecular Signatures of Alzheimer’s Disease

“…The phosphorylation is a type of posttranslational modification of proteins that regulates various aspects of their functionalities [60,61]. Protein phosphorylation plays a key role in cell signaling, gene expression, and differentiation [62,63].…”