| As one of the most important biological macromolecules,protein is an important component of cells.It participates in various important life activities,such as gene replication,transcription,translation and regulation,signal transduction,immune recognition,cell transportation,reaction catalysis,etc.However,in these important life activities,proteins do not function in the form of a single molecule,but perform their biological functions through interaction with other molecules,and protein-protein interaction plays an important role.Studying the interaction between proteins is an important method to study the structure and properties of molecular systems.It has been widely used in chemical engineering,biological medicine,material science and engineering,physics and other disciplines.An important step in understanding proteinprotein interaction is protein-protein docking,which can be defined as finding the binding conformation of two proteins through calculation,thus forming a stable complex with low energy.Therefore,the description,modeling,prediction and design of protein-protein interactions are very important for the theoretical research and practical application of proteomics.There are three key steps in protein-protein docking:(1)building the docking system,(2)using the scoring function to score the conformation,and(3)performing global search for the conformation according to the score.At present,most of the studies on protein-protein docking focus on the second step,and different scoring functions emerge in endlessly,ignoring the global conformation search part.In fact,the search sampling space in protein-protein docking is very large:thousands of protein frames,about 1016 conformational translation and rotation modes.It is obviously impossible to only traverse this size of space,and some existing algorithms,such as umbrella sampling and metadynamics,also have a huge amount of computation,and cannot quickly find the global optimal solution under the premise of ensuring accuracy.In biological processes such as protein-protein docking,the energy change is also continuous and smooth when the complex conformation changes continuously through a specific sequence.This thesis will study the sampling problem of protein-protein docking conformation change based on the CUR low-rank matrix decomposition technology.The first step is to select and compare different kinds of scoring functions,model the conformational changes of small proteins,and draw the energy surface to restore it using the optimization algorithm to verify the availability of the algorithm;Then,taking protein-protein docking as an example in a simple scenario,we use CUR sampling to sample the conformation,and compare and analyze the optimal conformation obtained with the results obtained by the mainstream protein-protein docking software to verify the correctness of the sampling method and the accuracy of the algorithm;Finally,the search process of the existing ab initio design protein-binding protein algorithm is improved,which greatly accelerates the protein-protein docking speed and promotes the protein research process while ensuring the accuracy. |
