Researchers use different modern technologies to combat such diseases and deep learning is one among them with faster prediction and achieves greater than accuracy.A var iety of repurposed drugs and investigational drugs have been identified in the past.Hundreds of clinical tr ials involving remdesivir, chloroquine, favipiravir, chloroquine, convalescent plasma, TCM and other interventions are planned or underway. Lancet. Lancet. J Biomol Struct Dyn. https: doi.org. Recent studies have determined that some diseases such as cancer, diabetes, and neurodegenerative diseases are caused by abnormal phosphorylation.Based on its potential applications in biological research and drug development, the largescale identification of phosphorylation sites has attracted interest.Existing wetlab technologies for targeting phosphorylation sites are overpriced and time consuming.Thus, computational algorithms that can efficiently accelerate the annotation of phosphorylation sites from massive protein sequences are needed.Numerous machine learningbased methods have been implemented for phosphorylation sites prediction.However, despite extensive efforts, existing computational approaches continue to have inadequate performance, particularly in terms of overall ACC, MCC, and AUC.The proposed technique expediently learns the protein representations from conjoint protein descriptors.Protein phosphorylation has significant functions, particularly in the regulation of diverse cellular processes in both prokaryotic and eukaryotic organisms and cell cycle control. It has been proposed that at least onethird of the cellular proteins in eukaryotic organisms are modified by phosphorylation and that of them are causative of multiple types of human diseases, especially cancer. Recent research has revealed that the study of kinases and their substrates are critical for understanding the signaling networks in cells and can aid the development of new treatments for diseases induced by signal irregularity, such as cancer. Therefore, the identification of phosphorylation sites may help reveal the molecular mechanisms of phosphorylationrelated biological processes.To date, copious computational methods have been developed for identification of phosphorylation sites.These methods can be categorized broadly into two groups, computational methods based on machine learning. The neighboring amino acids of phosphorylation residue may not individually identify that a specific site is activated; therefore, only discriminative patterns based methods are incompetent to distinguish phosphorylation sites. Despite the extensive progress achieved in the prediction of protein phosphorylation sites, existing algorithms have many shortcomings, and opportunity exists for improving prediction performance.One limitation of existing tools is their reliance on traditional shallow ML methods for the prediction of phosphorylation sites; these methods fail to learn the underlying biological features of phosphorylation sites and thus result in S.A second limitation is that existing feature extraction techniques are unable to adequately describe the biological properties of the protein modification sites.A robust feature optimization algorithm is indispensable to select the best discrimination feature subset for the final prediction model.Therefore, it is of utmost importance to develop a novel predictor that can abstract highlevel patterns from the sequences to produce accurate prediction results with a low error rate.Deep learning is considered a powerful solution for such problems; it entails a model architecture composed of multiple layers of neural networks that can extract highlevel abstractions from data automatically.Deep learning approaches have demonstrated outstanding results compared with popular shallow ML algorithms in several research areas, such as speech recognition.