Visualanalysis Of Pocket Dynamic Patterns Based On Implicit Characterization Of Inner Wallatoms

Protein pockets are an essential component of protein molecules.They play a crucial role in regulating protein function,and studying pockets forms the foundation for understanding protein functionality.Since pockets undergo changes as protein molecules move,their intrinsic dynamic properties are of significant importance to protein functional mechanisms and can aid in drug discovery and development.How to accurately identify and characterize dynamic pockets,as well as exploring and analyzing their motion pattern,is crucial to computer-aided drug design.This paper presents a method based on implicit representation to explore and analyze the motion patterns of dynamic pockets,taking into consideration their high dimensionality,spatiotemporal characteristics,and correlation.Visualization is employed as a means to explore and analyze the motion patterns of pockets.Firstly,in order to better represent dynamic pocket data,this paper proposes a dynamic pocket characterization method based on the amino acid atoms of the pocket inner wall.This method processes the protein data obtained through molecular dynamics simulation,identifies different pockets,and extracts the sequence of amino acid atoms along the inner wall of the pocket to obtain a dynamic pocket feature vector with high dimensionality,spatiotemporal characteristics,and correlation.Secondly,in order to quickly detect the similarity and difference between pockets and locate the important areas of protein molecule dynamic motion,this thesis proposes a Jaccard similarity coefficient-based computational method.This method solves problems such as mismatches in the time length and temporal correlation of two dynamic pocket sequences,as well as dimension mismatches and correlation shifts between two static pockets.Finally,to assist domain experts in analyzing the stability of dynamic pockets and the motion patterns of pockets in the drug design process,this thesis proposes an interactive exploration and visualization platform based on“experts in the loop”.The spatiotemporal changes of dynamic pockets are presented in a visual form,enabling domain experts to better understand the motion patterns of protein dynamic pockets and identify similarities and differences between different pockets.The research work presented in this paper proposes a new pocket characterization method and designs a visual analysis system for exploring dynamic pockets from different perspectives,demonstrating the effectiveness and practicality of the proposed method.