A Gene Regulatory Network Construction Method Based on Bidirectional xgBoost

Abstract

The invention discloses a method for constructing a gene regulation network based on bidirectional XGBoost. According to the characteristics of time series gene expression data, the relationship between the expression value of a gene at time t and other genes before and after time t is analyzed, and local-in and local ‑out two-way model; in a single model, the construction of the gene regulatory network is divided into multiple regression equations, fully considering the situation of high-order time lag, using XGBoost to evaluate the weight of the features in a single regression equation, and finally obtain the ranking of the regulatory relationship ; Use the normalized residual to reweight the features in each independent regression equation twice, and integrate the weight of the two-way model on the same pair of regulatory relationships as the final scoring value for each pair of gene regulatory relationships. The invention is applicable to the construction of gene regulatory networks of different scales under time series gene expression data, and can effectively improve the accuracy of network inference.

Description

technical field [0001] The invention relates to the field of bioinformatics, and relates to a method for constructing complex biological networks. Background technique [0002] The life activities of organisms are based on the interaction of a large number of biomolecules, including the signal transduction between cells, the interaction between proteins, and the mutual influence of genes. Many complex connections can be abstractly expressed in the form of biological networks. Among many biological networks, gene regulatory network represents a mechanism to control gene expression levels in organisms. According to the gene regulatory network, isolated studies of an isolated gene often cannot reflect its complex changes, and the expression level of a gene is affected by the expression content of other genes, including promotion and inhibition relationships. Gene regulation occurs in multiple stages such as biological transcription and translation, and is affected by specific ...

AI Technical Summary

Problems solved by technology

[0007] Although there are a large number of gene regulatory network construction methods, these methods are limited by the network scale and computational model, and there are still many deficiencies in construction ability and accuracy.

Images