Lysine acetylation site prediction method based on modular dense convolutional network

Abstract

The invention discloses a lysine acetylation site prediction method based on a modular dense convolutional network, and the method comprises the steps: introducing the structural characteristics of aprotein, and combining the structural characteristics with the original sequence of the protein and the physicochemical attributes of amino acids to construct a site feature space; adopting a modulardense convolutional network to capture characteristic information of different hierarchies, and reducing information loss and information crosstalk in the characteristic learning process; and introducing a compression-excitation layer to evaluate the importance of different characteristics, so that the abstraction capability of the network is improved, and the potential lysine acetylation sites are identified. The method can effectively solve the problem that an existing method only considers protein sequence level information and is low in characteristic learning efficiency, potential lysineacetylation sites are more accurately predicted, the verification cost of the lysine acetylation sites is reduced, and the research efficiency of lysine acetylation modification is improved.

Description

technical field [0001] The invention relates to the field of lysine acetylation site prediction research and analysis, in particular to a lysine acetylation site prediction method based on a modular dense convolutional network. Background technique [0002] Lysine acetylation is a conserved protein post-translational modification, which is closely related to a variety of metabolic diseases. Therefore, the identification of lysine acetylation sites is of great significance for the research on the treatment of metabolic diseases. Protein structural properties contain highly useful structural information, which provides a strong basis for the identification of protein post-translational modifications; in the process of feature learning, the information between different levels of features is complementary, and paying attention to the information of different levels of features can effectively improve the quality of features. Existing deep learning methods use information at the...

AI Technical Summary

Problems solved by technology

Existing deep learning methods use information at the protein sequence level as input, without considering the structural properties of proteins; only high-level features are considered during feature extraction, resulting in serious loss of information, which in turn reduces the prediction results.

Images