Global Alignment Of Protein Structure In Cryo-EM

Proteins are the main working molecules involved in various cellular processes within cells.In order to understand the function of proteins,people have made great efforts to solve the three-dimensional structure of proteins by experimental methods,including X-ray crystallography,nuclear magnetic resonance technology,cryo-electron microscopy,and so on.In recent years,the technological revolution has greatly improved the potential of cryo-electron microscopy to produce macromolecular structures,making cryo-electron microscopy increasingly the mainstream technology of structural biology.The structures of various biological macromolecules such as proteins are measured by cryo-electron microscopy(Cryo-EM)and stored in the electron microscopy database(EMDB)in the form of electron microscopy maps(EM maps).The development of cryo-electron microscopy technology has led to a significant accumulation of EM maps in EMDB.As of February 2023,the database currently has over 24795 maps.The extensive production of EM maps has enriched the types of proteins in the database,and technological advancements have also increased the proportion of high-resolution EM images in EMDB,providing more assistance in analyzing and studying the structure and function of proteins.In order to utilize the rapidly accumulated EM map,it is necessary to analyze the structural information in EM map.In the process of analyzing the protein structure information in EM map,the alignment of protein structure in the EM map is an essential step,which is also the alignment of the EM map.Alignment involves many important processes in analyzing EM maps,including fitting the rigid body structure of atomic models,understanding the similarities and differences of different functional states,database search and so on,which have important practical significance.Most of the existing methods of protein structure alignment are aimed at the models in the protein three-dimensional structure database(PDB),while the methods of protein structure alignment in EM map are relatively limited,and these methods also have some limitations.Therefore,this paper designs a new algorithm flow to solve the problem of protein structure alignment in EM map,and constructs the similarity evaluation indicators of structural alignment to achieve the similarity retrieval of the EM map database.In the design of the algorithm process,it is mainly divided into two stages:converting the structure in EM map into a point cloud model and implementing rigid registration of the point cloud model.In the first stage,the algorithm process for converting EM maps into point clouds was designed to approximate the structure in EM map into a point cloud model.The experiments showed that the obtained models can effectively approximate the structure in EM map.In the second stage,a point cloud registration method that integrates 3DSIFT and 3DSC feature descriptor was designed.Using the point cloud model obtained in the previous stage as the experimental dataset,comparative experiments were conducted with other point cloud registration methods to verify that the registration accuracy of this method is higher.Finally,the two stages are merged to form the overall workflow of protein structure comparison in EM map.In the overall evaluation of the algorithm,the current mainstream evaluation indicators are used to evaluate this paper and other methods of protein structure alignment in EM map,which verifies that this paper has more advantages in the accuracy of structural alignment.Using the algorithm proposed in this paper to achieve similarity retrieval in EM map datasets,and comparing it with current mainstream similarity retrieval methods,it is verified that this paper has a higher accuracy in similarity search in EM map datasets.