Comparison Of 3D Structures Generated by AlphaFold2 to Experimental Structures In Oncogenic Proteins

Abstract

AlphaFold2 is a machine-learning algorithm that can predict the 3D structure of proteins. 3D protein structures are essential for understanding the function of oncogenic proteins, which can potentially cause cancer. In this study, we compared the structure of 26 oncogenic proteins found experimentally and computationally using AlphaFold2. We used RMSD values to measure how well the AlphaFold2 model fit the experimentally derived protein structures. RMSD values for the oncogenic proteins ranged from 0.204 Å to 1.980 Å with an average of 0.633 Å, showing that AlphaFold2 was a promising tool for predicting the 3D structure of oncogenic proteins. However, we noted that AlphaFold2 has limitations in predicting the structure of highly disordered proteins, proteins with multiple conformations, mutated and artificial proteins, and proteins that are not well-studied experimentally. Our study suggests that AlphaFold2 could be used to identify new targets for cancer treatment and design drugs that can fit into the binding sites of oncogenic proteins. We also hypothesize that AlphaFold2 could be improved by increasing the amount of data available for training, improving the resolution of current data, and using it in conjunction with other protein structure models. We believe that AlphaFold2 is a powerful tool for predicting the structure of oncogenic proteins. Despite the current limitations, we are optimistic that future research will improve AlphaFold2, finding use in cancer research.