Investigation Of Interaction Function Of Mutant Transmembrane Proteins Based On Binding Domain Prediction

As the main drug targets of transmembrane proteins,it is of great significance to study the effect of mutations on their interactive functions for drug development,drug efficacy and other problems.Protein structure determines its function,so this study is highly relevant to protein structure information.However,due to the limited current experimental technology,the crystal structure of transmembrane proteins measured before and after mutation is seriously missing,and the lack of exploration of binding domains closely related to interaction function limits relevant studies.The appearance of Alpha Fold2 protein structure prediction tool and the proof of the common structure of binding domain provide favorable conditions for the exploration of this problem.Therefore,in this paper,Alpha Fold2 was used to obtain the three-dimensional structure of the protein before and after mutation,and based on the optimized and improved binding domain model,the interactive functional changes of transmembrane protein mutation were explored and studied.Firstly,this paper demonstrates the outstanding performance of Alpha Fold2 on transmembrane proteins through a large number of statistical results and data analysis,and its average TM-score precision value can reach 0.85,which can be used as a tool for structure acquisition in this paper.Secondly,from the perspective of structure,a binding domain prediction model based on Res Net framework fusion graph convolutional neural network idea is designed,which optimizes and improves the effect of binding domain prediction,reaching0.58 on the important index MCC,providing the corresponding binding domain basic model for mutation research.The difference between this binding domain and the conventional concept of binding site region lies in the emphasis on structural features.It refers to the common spatial region of interaction in three-dimensional structure,which can reflect the relevant situation of interaction functions.Finally,a global statistical analysis of the changes in the interaction function of transmembrane protein mutation was carried out,and a local dichotomy model was explored based on the multi-layer perceptron and residual network respectively,which obtained the basic classification effect,and the accuracy and other indicators reached more than 0.65.Finally,the following applications can be realized through the research work of this paper: Given a mutated transmembrane protein,Alpha Fold2 was used to obtain the three-dimensional structure of the protein before and after mutation,the optimized binding domain model was used to obtain the relevant features of binding domain,and the binary exploration model was used to obtain the important trend of functional changes caused by mutation,so as to realize the study on the interactive function of mutant transmembrane proteins based on binding domain prediction.In conclusion,the work in this paper provides a basis for the application value and scenarios of Alpha Fold2 in transmembrane protein-related studies,optimizes and improves the solution to the problem of transmembrane protein binding domain prediction,and analyzes and explores the changes in transmembrane protein mutation interaction function from the perspective of binding domain.This will provide validation basis and method support for many experimental studies related to transmembrane proteins and their mutations,such as drug target action,drug development and drug efficacy.