Construction Of A Gene Regulatory Network For Brassica Napus L Based On Machine Learning

Brassica napus L.is an allotetraploid crop,and its gene regulatory network is complex,which affects the mining and analysis of the gene functions of Brassica napus.Gene regulatory network is a directed network that with the function of expressing the interaction relationship between genes.We analyzed the Brassica napus transcriptome data through machine learning,and then constructed genome-wide gene regulatory networks(GRNs).Using this network,the transcription factors involved in the auxin synthesis pathway of Brassica napus,and the downstream target genes of the auxin response factor ARFs protein family were preliminarily mined.The relevant results are as follows:(1)Using the transcriptome data,DAP-seq and Ch IP-seq data of various tissues of Brassica napus and Arabidopsis thaliana in the NCBI public database,the networks of Arabidopsis thaliana and Brassica napus were constructed by GENIE3 based Random Forest and Extra-tree methods,respectively.Then,the Arabidopsis thaliana network was used to correct the Brassica napus network,and finally six Brassica napus GRNs were obtained.The results of network sensitivity test for six Brassica napus GRNs showed that71% of the optimized Brassica napus network sensitivity was improved.Taking auxin ARFs protein transcription factors as an example,Receiver Operating Characteristic(ROC)analysis showed that 92.59% of the AUC values of ARFs family networks were higher than0.5.Furthermore,it was found that all GRNs network conformed to the power function distribution,which indicated this network conformed to the scale-free network analysis and can be screened and utilized by using Degree centrality and other indicators.(2)Auxin is a vital plant hormone,using the rapeseed GRNs network,we mined the transcription factors involved in the regulation of auxin synthesis pathways.There were a total of 75 transcription factors with a Degree centrality greater than 40 in the network.The results of gene function annotation showed that these transcription factors were related to hormones.Among them,WRKY family accounted for 17.27%,indicating that these WRKY transcription factor members may be involved in the regulation of rapeseed auxin synthesis.At the same time,we analyzed the downstream network regulated by the ARFs family of transcription factors in the auxin signaling pathway,and the results showed that the Degree centrality of ARF2 in Brassica napus accounted for 41% of all ARFs family genes,which was significantly higher than other members.GO annotations results showed that genes in cell cycle,organ development,and auxin related pathways were significantly enriched.The downstream target genes differed significantly among homologous copies,but there were also target genes that were co-regulated by multiple copies.In addition,the results of chromosome bias analysis showed that the target genes of 3 genes including ARF1 were significantly biased towards C subgenome,while the target genes of 4 genes including ARF3 were significantly biased towards A subgenome.We constructed rapeseed gene regulatory network through machine learning.Through this network we preliminarily mined transcription factors that regulate auxin synthesis and downstream target genes of the ARFs protein family.The study suggested GRNs was helpful for predicting and mining the regulatory factors of known pathways and laying a foundation for the research of functional genomics in Brassica napus.