mCNN-glucose: Identifying families of glucose transporters using a deep convolutional neural network based on multiple-scanning windows

Glucose transporters are essential carrier proteins that function on the phospholipid bilayer to facilitate glucose diffusion across cell membranes. The transporters play many physiological and pathological roles in addition to absorption and metabolism of fructose in food and the pathogenesis of gastrointestinal diseases. These carrier proteins play an important role in diseases of the nervous system, cardiovascular system, digestive system, and urinary system. These essential transporters have been extensively studied as potential therapeutic targets for cancers such as pancreatic, prostate, and hepatocellular carcinoma, which serve as diagnostic and prognostic indicators. The method uses position-specific scoring metrics (PSSM) with multiple-scanning windows-based convolutional neural networks to classify glucose transport proteins based on their functional significance and crucial role in therapy. Convolutional neural networks with multiple window scanning are employed to capture biologically meaningful, significant, and meaningful features from PSSM evolutionary profiles. Our proposed Method obtained Matthews correlation coefficients (MCC) of 0.99, Accuracy (AC) of 99.46, for Glucose facilitative transporters (GLUT), 0.99, 99.46, for Sodium Coupled glucose transporters (SGLT), and 0.92, and 97.3 for Sugars will eventually be exported transporters (SWEET) respectively. This study shows significantly higher performance than our previous study, which could be used to accurately classify novel glucose transporters.