AUTHOR=Sikander Rahu , Arif Muhammad , Ghulam Ali , Worachartcheewan Apilak , Thafar Maha A. , Habib Shabana 

TITLE=Identification of the ubiquitin–proteasome pathway domain by hyperparameter optimization based on a 2D convolutional neural network

JOURNAL=Frontiers in Genetics

VOLUME=Volume 13 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/genetics/articles/10.3389/fgene.2022.851688

DOI=10.3389/fgene.2022.851688

ISSN=1664-8021

ABSTRACT=The major mechanism of proteolysis in the cytosol and nucleus is the ubiquitin–proteasome pathway (UPP). The highly controlled UPP has an effect on a wide range of cellular processes and substrates, and flaws in the system can lead to the pathogenesis of a number of serious human diseases. Almost all cellular processes are regulated by the UPP, and deep learning algorithms are difficult to optimize on hyperparameters. In this paper, this solution used a genetic algorithm. Accuracy and validation of the two different but closely related stages were examined. An algorithm with a two-dimensional convolution and a fully connected layer was evaluated to achieve three separate models, and a three-model approach was used. The data in the 2DCNN-UPP dataset were used for cross-validation, and an independent process diagnosis dataset was also incorporated. Evidence showed that the method can save time on training. In the case where experimentally validated ubiquitination sites emerged, we must devise a proteomics-based predictor of ubiquitination. We proposed 2D-CNN-UPP, a deep learning model, which used convolutional neural networks to predict the results. Four approaches were used in the sequences, and the physical properties were combined. When using 10-fold cross-validation, 2D-CNN-UPP obtained an AUC (ROC) value of 0.801 predicted score. We analyzed the relationship between UPP protein and the non-UPP protein predicted score. This study used quantitative techniques and obtained 0.862% accuracy, 0.921% sensitivity, 0.803% specificity, and 0.730% MCC predicted score, which indicated that the performance of the model was remarkable. By using our proposed computational method, this research could effectively analyze UPP proteins, and our research work might improve computational biological research. Therefore, we could utilize the latest features in our model framework and DDE-based protein structure features for the prediction of protein structure, functions, and different molecules, such as DNA and RNA.