Multi-objective evolutionary algorithm-based large-scale protein function module identification method

Abstract

The invention discloses a multi-objective evolutionary algorithm-based large-scale protein function module identification method. The method comprises the steps of defining a protein network representation, and identifying core protein nodes in a protein network; establishing a sub-population based on the core protein nodes, and performing initialization operation on individuals in the sub-population; S3, performing crossover mutation operation on the individuals subjected to the initialization operation to obtain new individuals; performing crossover mutation operation on the individuals in anew sub-population to obtain new individuals, calculating module degrees of the new individuals, searching for the new individual with the maximum module degree, and recording the maximum module degree; and performing gaining according to the maximum module degree, and combining protein modules with the overlapping degree exceeding a preset value in the protein modules. The search capability of amulti-objective evolutionary algorithm is improved; the algorithm pays more attention to and surrounds the core nodes to perform protein module search; and the mined protein modules are more exquisite and accurate, so that the validity of protein module mining is ensured.

Description

technical field [0001] The invention relates to the technical field of protein network functional module identification, in particular to a large-scale protein functional module identification method based on a multi-objective evolutionary algorithm. Background technique [0002] The study of protein functional modules is of great significance for understanding the organizational structure of biological systems. The protein modules in organisms are complex and diverse, and in the post-gene era, the scale of protein networks is generally on the order of one hundred thousand million. How to quickly and effectively identify various modules with biological functions is a key scientific problem in proteomics. However, the existing schemes cannot effectively solve the problem of identifying protein functional modules. Some of these schemes obtain single results, and some schemes require a lot of computing time, which cannot provide users with various practical analysis. [0003]...

AI Technical Summary

Problems solved by technology

In the past, such traditional detection methods had limitations in detection cost, time and quality, and it was difficult to meet the actual needs of life science research in the post-gene era; the second is to use the network topology of protein interaction networks to discover functional modules, and the research found that PPI network Those closely related protein regions usually correspond to functional modules, use the existing protein interaction database to construct protein interaction, and then use machine learning and data mining algorithms and ideas to identify protein functional modules

This type of method can quickly meet the needs of users, but protein networks form protein modules with a wide variety of functions at different time and different spatial stages. need

[0004] Mining protein functional modules based on multi-objective evolutionary algorithm can better dig out richer protein modules. This type of algorithm does not need to know the number of network communities, and finally obtains a set of solution sets, which can provide more information for decision makers. However, the multi-objective evolutionary algorithm has high computational complexity for large-scale protein networks, and cannot realize fast and efficient mining of protein modules.

Images