| MicroRNAs(miRNAs)is a kind of endogenous non-coding RNA,which participates in the physiological and biochemical process of crop growth and development by cutting and inhibiting the expression of target genes.In rice,some miRNAs have been reported to be important non-coding regulators involved in endosperm development.However,the pathways and regulatory pathways involved in it still need to be further studied.In addition,there are many miRNA prediction and analysis software,so how to select the suitable miRNA analysis software for rice is also a problem.In this study,a number of commonly used miRNA analysis software are selected to comprehensively evaluate the reliability of their prediction of known miRNA.Then the software suitable for rice miRNA research was selected to analyze the miRNA related to endosperm development.Finally,the potential biological process of endosperm development regulated by miRNA was analyzed by means of co-regulatory network.We selected 8 kinds of commonly used miRNA software to predict the known miRNA of high-pass sequencing data of Arabidopsis thaliana,rice,maize and wheat.In this study,combined with the miRNAs annotation information of database miRBase and sRNAanno,using random forest,BLAST and ROC(Receiver Operating Characteristic)curve algorithms,horizontal(intra-species)and vertical(inter-species)were used to evaluate the sensitivity,accuracy,specificity,computing time and RAM utilization of each software.The results show that:(1)the accuracy of sRNAbench is the highest,the sensitivity of miRExpress;(2)miRDeep-P2 is the highest and the prediction time of known miRNA is the shortest,and(3)the memory utilization of Mirkwood is the highest.On the basis of all the evaluation results,we think that sRNAbench and miRExpress are the two best softwares to predict the known miRNA of rice.Based on the above results,we selected the sRNA-seq data of rice endosperm related to rice endosperm and other rice tissues from 28 public data,and used sRNAbench and miRExpress software for known miRNA analysis.Difference table analysis showed that 60 endosperm-specific expressions were up-regulated by miRNA and 85 were down-regulated by the two softwares.In addition,we collected 78 rice endosperm-related RNA-seq data for ab initio splicing of transcriptome and construction of endosperm tissue-specific co-expression network,and finally obtained 551 potential lncRNA and co-regulatory network containing 50937 nodes(nodes)and 1789707 co-expression relationship(relationships).Then,based on principal component analysis,cluster analysis and pathway enrichment analysis,we selected the RNA-Seq data of 41 rice endosperm related samples and 32 other rice tissues to screen the endosperm-specific expression of miRNA and its target gene pairs.The results showed that 83 lncRNA were down-regulated,31 were up-regulated and 465 genes were down-regulated.Among them,osamiRNA1428a-3p has been reported to target LOC_Os03g17980 gene,that is,osk3 gene,which regulates endosperm by affecting starch accumulation in the early stage of endosperm development.Osa-miRNA398a has been reported to target LOC_Os07g46990 gene,that is,OsCSD gene,which regulates grain weight and participates in the growth and development of endosperm.Combining the target gene co-regulatory network,functional enrichment and literature review,we also predicted the known miRNA that miR166a-3p,miR171i-5p and miR408-5p may affect endosperm development.However,further biological experiments are needed to prove whether these three miRNAs can really affect the endosperm development of rice.In this study,software evaluation,transcriptome data collection and calculation,co-regulatory network construction and modular function prediction were integrated to mine the non-coding RNA regulating rice endosperm development and its potential participation in biological processes at the genome-wide level.It is intended to provide data and theoretical support for the cultivation of good quality rice. |
PDF Full Text Request