Research On Algorithms Of Gene Regulatory Network Reconstruction Based On Single Cell Sequencing Data

A Gene Regulatory Network(Gene Regulatory Network)embodies the interaction and mutual influence between genes.The construction of a gene regulatory network helps to reveal the specific role of each gene in a specific life process or disease development process,and is helpful to understand the function of each gene.The recently proposed transcription dynamics models based on single-cell sequencing data can better describe the dynamic process of gene expression.Most of the previous gene regulatory network reconstruction algorithms did not use single-cell sequencing data,and a few algorithms that used these data did not use it.Using the dynamics models this paper proposes new gene regulatory network reconstruction algorithms GRN-VM and GRN-ML based on single-cell sequencing data.The main contributions are as follows:(1)Gene Regulatory Network Reconstruction Algorithm Based on RNA Velocity Modeling(GRN-VM).The RNA velocity model has been proven to be an effective transcriptional dynamics model.It can restore the direction of cell differentiation in the cell lineage to a certain extent.The changes in the transcriptome during the differentiation process are determined by complex gene regulation,so RNA velocity may imply information about the interaction between genes.For single-cell transcriptome sequencing data,the original transcription dynamics model did not consider the dynamic information of the m RNA splicing rate.GRN-VM proposed a new RNA velocity model,which introduces the m RNA splicing rate as an extra parameter And analyzes the unspliced m RNA and spliced m RNA in cell to calculate the RNA velocity of gene at current moment.GRN-VM uses the ordinary differential equation(ODE)to establish the RNA velocity model,estimates the transcription rate,splicing rate and degradation rate of genes in the RNA velocity model,and then pass the RNA speed model with the single-cell transcriptome data in a specific Combining methods.Finally,GRN-VM reconstructs a gene regulatory network using random forest.Extensive experimental results based on simulated data show that the GRN-VM algorithm has better performance than the recent similar algorithm dyn GENIE3 algorithm.(2)Gene Regulatory Network Reconstruction Algorithm Based on Metabolic Labeling Dynamics Modeling(GRN-ML).The latest single-cell sequencing technology can perform metabolic labeling experiments at single-cell resolution,which record the dynamic process of the changes of the labeled m RNAs over time.In order to deal with the challenge that it is difficult to fit an RNA velocity model in steady-state cells,GRN-ML introduces a metabolic labeling dynamic model,which uses metabolic labeling data to calculate the true m RNA transcription rate,splicing rate and degradation rate.GRN-ML integrates these parameters into the dynamics model to reconstruct the gene regulatory network.The results of experiments based on simulation data show that the performance of GRN-ML is significantly improved compared to the previous gene regulatory network reconstruction algorithms...