The Swarm Intelligence Optimization Algorithm Identifies Key Proteins And Their Complexes

With the rapid development of bioinformatics,a large number of available protein data has been generated and used in related research.One of the hotspots worth studying in the field of bioinformatics at present is to effectively analyze the protein structure and function by using computational methods to enhance the understanding of the life process of organisms.In particular,the discovery and detection of essential proteins and protein complexes in protein-protein interaction networks is of great significance for revealing the function and organization of organisms.In recent years,the rise and maturlty of swarm intelligence optimization algorithms has provided a new perspective for essential protein prediction and protein complex detection.In the essential protein prediction,this thesis mainly uses the swarm intelligence optimization algorithm to design the corresponding relationship between it and the essential protein identification and construct the corresponding identification nodel,or proposes an effective feature selection strategy based on the swarm intelligence optimization mechanism to judge whether proteins are essential to improve recognition accuracy.Considering the close relationship between essential proteins and protein complexes,in the protein complex detection,this thesis mainly combines essential proteins to construct the protein complex detection model based on the swarm intelligence optimization process to improve the recognition performance of protein complexes.The main work of this thesis is as follows:First of all,essential protein recognition model based on the swarm intelligence optimization process.Aiming at the defect of low prediction accuracy of essential proteins,an improved flower pollination algorithm was proposed to identify essential proteins.By simulating the flower pollination process in the flower pollination algorithm,the corresponding relationship between this process and essential protein recognition was designed to construct a novel essential protein recognition model for flower pollination optimization.At the same time,the topological and biological properties of protein-protein interaction network were fully integrated.The experimental results show that the prediction accuracy of the improved algorithm is better than that of the classical essential protein recognition algorithms.Secondly,feature selection strategy based on swarm intelligence optimization predicts essential proteins.Aiming at most of the essential protein identification methods are based on a few features to identify essential proteins,there is a lack of analysis of multiple topological and biological features,a feature selection strategy based on the flower pollination optimization was proposed to predict the essentiality of protein.Collecting various topological and biological features related to protein essentiality,and combing the flower pollination algorithm with elite search mechanism to select the optimal feature subset to determine whether each protein is an essential protein.Experimental results show that the new feature selection strategy has better performance than the classical feature selection methods.Thirdly,protein complex detection algorithm based on swarm intelligence optimization mechanism.Aiming at the flaw in traditional protein complex detection,and considering the close relationship between essential proteins and protein complexes,the protein complex detection algorithms based on flower pollination and moth-flame mechanism were designed respectively.In the protein complex detection algorithm based on flower pollination mechanism,by considering data sources including essential protein information,location information,function information and topology information of proteins,the multi-relation reconstructed dynamic protein networks were established,and then combining the structure of protein complex to find the potential cores of protein complexes in these networks,and finally the attachment proteins were added to the corresponding cores to form protein complexes,the experimental results show that the newly proposed protein complex detection algorithm can detect protein complexes more accurately than the classical algorithms.In the protein complex detection algorithm based on moth-flame mechanism,the reliable dynamic protein networks were constructed from different angles,then the layer-by-layer concept was introduced and essential protein information was integrated to find the cores of protein complexes known as the flames,so that the moths flew around the flames in a spiral form to generate the protein complexes,and finally filter operation was used to refine the final predicted protein complexes,the experimental results show that the newly designed protein complex detection algorithm can detect protein complexes with higher accuracy than the classical algorithms.