Research On Functional Module Detection In Protein-protein Interaction Networks Based On Bat Algorithm

Protein-protein interaction network(PPIN)is composed of the interactions of all proteins in a living organism,with the characteristics of a highly complex network.Detection of functional modules in PPIN is an important issue in proteomics research.It is not only of great significance to reveal the functions of proteins and understand the tissue structure of biological systems,but also provides theoretical basis for the prediction of diseases and the discovery of drug targets in the future.In order to accurately identify the PPIN functional modules,a large number of computing methods are rapidly emerging.Among them,applying the idea of swarm intelligence to detect functional modules has become a research hotspot in the field.This topic explores a novel swarm intelligence algorithm,bat algorithm,to solve the PPIN functional module detection problem,and carried out the following works from two aspects:(1)Bat algorithm is a swarm intelligence algorithm with strong global search ability.In this paper,a method based on bat algorithm for functional modules detection(BA-FMD)in PPIN is proposed to achieve good results in this field.Firstly,the method combines the topology information and function information between the nodes in the PPIN as their similarity measure,and uses the similarity measure to randomly initialize the bat population in the search space.The position of each bat represents a candidate functional module division.Then,we design the directional local disturbance,random disturbance,adaptive variation based on distance and frequency,and natural selection for the random optimization of solutions in the process of the population optimization.Finally,the optimal or sub-optimal functional module division is further modified by two post-processing operations including merging and filtering.The BA-FMD algorithm is compared with six classical algorithms with different mechanisms on five PPIN datasets having different scale.The results show that the proposed algorithm has certain competitiveness in detection quality.(2)Aiming at the shortcomings of BA-FMD algorithm that ignores information interaction within the population and easily falls into local optimum,this paper fuse immune genetic algorithm focusing on information sharing into BA-FMD algorithm,and propose a method based on fusion of bat algorithm and immune genetic algorithm for functional modules detection(BAIGA-FMD)in PPIN.In the framework of the bat algorithm,the algorithm adds the main idea of immune genetic algorithm in the process of evolution.Firstly,the position of each bat individual is corresponding to an antibody,and the fitness value is mapped to the affinity measurement between antibody and antigen,and calculating the concentration value of each antibody.Then,the corresponding antibodies are extracted by combining affinity and concentration,and mutation,crossover and reorganization are carried out among these individuals,which increase the communication among individuals,increase the diversity of the population,and prevent the population from falling into local optimum too early at the same time.Finally,the prior knowledge is extracted by immune operator according to the optimal solutions of past and contemporary generations,which enhances the learning ability of the population and makes the algorithm obtain better solutions.The experimental results on five PPIN datasets show that BAIGA-FMD algorithm has a certain improvement than the original algorithm in many indicators,and the comparison with other six classical algorithms further shows the effectiveness of the new algorithm.