Introduction: With the advancement of RNA-seq technology and machine learning, training large-scale RNA-seq data from databases with machine learning models can generally identify genes with important regulatory roles that were previously missed by standard linear analytic methodologies. Finding tissue-specific genes could improve our comprehension of the relationship between tissues and genes. However, few machine learning models for transcriptome data have been deployed and compared to identify tissue-specific genes, particularly for plants. Methods: In this study, an expression matrix was processed with linear models (Limma), machine learning models (LightGBM), and deep learning models (CNN) with information gain and the SHAP strategy based on 1,548 maize multi-tissue RNA-seq data obtained from a public database to identify tissue-specific genes. In terms of validation, V-measure values were computed based on k-means clustering of the gene sets to evaluate their technical complementarity. Furthermore, GO analysis and literature retrieval were used to validate the functions and research status of these genes. Results: Based on clustering validation, the convolutional neural network outperformed others with higher V-measure values as 0.647, indicating that its gene set could cover as many specific properties of various tissues as possible, whereas LightGBM discovered key transcription factors. The combination of three gene sets produced 78 core tissue-specific genes that had previously been shown in the literature to be biologically significant. Discussion: Different tissue-specific gene sets were identified due to the distinct interpretation strategy for machine learning models and researchers may use multiple methodologies and strategies for tissue-specific gene sets based on their goals, types of data, and computational resources. This study provided comparative insight for large-scale data mining of transcriptome datasets, shedding light on resolving high dimensions and bias difficulties in bioinformatics data processing.

Background: Coronavirus (CoV) is an emerging human pathogen causing severe acute respiratory syndrome (SARS) around the world. Earlier identification of biomarkers for SARS can facilitate detection and reduce the mortality rate of the disease. Thus, by integrated network analysis and structural modeling approach, we aimed to explore the potential drug targets and the candidate drugs for coronavirus medicated SARS. Methods: Differentially expression (DE) analysis of CoV infected host genes (HGs) expression profiles was conducted by using the Limma. Highly integrated DE-CoV-HGs were selected to construct the protein-protein interaction (PPI) network.  Results: Using the Walktrap algorithm highly interconnected modules include module 1 (202 nodes); module 2 (126 nodes) and module 3 (121 nodes) modules were retrieved from the PPI network. MYC, HDAC9, NCOA3, CEBPB, VEGFA, BCL3, SMAD3, SMURF1, KLHL12, CBL, ERBB4, and CRKL were identified as potential drug targets (PDTs), which are highly expressed in the human respiratory system after CoV infection. Functional terms growth factor receptor binding, c-type lectin receptor signaling, interleukin-1 mediated signaling, TAP dependent antigen processing and presentation of peptide antigen via MHC class I, stimulatory T cell receptor signaling, and innate immune response signaling pathways, signal transduction and cytokine immune signaling pathways were enriched in the modules. Protein-protein docking results demonstrated the strong binding affinity (-314.57 kcal/mol) of the ERBB4-3cLpro complex which was selected as a drug target. In addition, molecular dynamics simulations indicated the structural stability and flexibility of the ERBB4-3cLpro complex. Further, Wortmannin was proposed as a candidate drug to ERBB4 to control SARS-CoV-2 pathogenesis through inhibit receptor tyrosine kinase-dependent macropinocytosis, MAPK signaling, and NF-kb singling pathways that regulate host cell entry, replication, and modulation of the host immune system. Conclusion: We conclude that CoV drug target \"ERBB4\" and candidate drug \"Wortmannin\" provide insights on the possible personalized therapeutics for emerging COVID-19.

Human mesenchymal stem cells (hMSCs) have the capacity to differentiate into fabricate cartilage, muscle, marrow stroma, tendon/ligament, fat, and other connective tissues, providing a potential source for tissue regeneration. The aim of this study was to find the key transcription factors (TFs), which regulated osteogenic differentiation of hMSCs. In this study, three methods were performed to find the key TFs, which included enrichment analysis, direct impact value and indirect impact value. We used the patient and public involvements (PPI) network to integrate the results of the above methods for analysis. Then, we compared the osteoblast data to the control group on days 1, 3 and 7. Finally, we found the combination of the optimal and vital 30 TFs related to osteogenic differentiation. TFs FOS, SOX9 and EP300 were commonly expressed in 3 different days in the osteogenic lineages and presented in the PPI network at relatively high degrees. Moreover, TFs CREBBP, ESR1 and EGR1 also presented high effects on the 1st, 3rd and 7th day. The constructed network gives us a more comprehensive understanding of the mechanism of osteogenesis of hMSCs.

Microarray data have vastly accumulated in the past two decades. Due to the high-throughput characteristic of microarray techniques, it has transformed biological studies from specific genes to transcriptome level, and deeply boosted many fields of biological studies. While microarray offers great advantages for expression profiling, on the other hand it faces a lot challenges for computational analysis. In this chapter, we demonstrate how to perform standard analysis including data preprocessing, quality assessment, differential expression analysis, and general downstream analyses.

Recurrent aphthous stomatitis (RAS) represents the most common chronic oral diseases with the prevalence ranges from 5% to 25% for different populations. Its pathogenesis remains poorly understood, which limits the development of effective drugs and treatment methods. In this study, we conducted systemic bioinformatics analysis of gene expression profiles from the Gene Expression Omnibus (GEO) to identify potential drug targets for RAS. We firstly downloaded the gene microarray datasets with the accession number of GSE37265 from GEO and performed robust multi-array (RMA) normalization with affy R programming package. Secondly, differential expression genes (DEGs) in RAS samples compared with control samples were identified based on limma package. Enriched gene ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways of DEGs were obtained through the Database for Annotation, Visualization and Integrated Discovery (DAVID). Finally, protein-protein interaction (PPI) network was constructed based on the combination of HPRD and BioGrid databases. What\'s more, we identified modules of PPI network through MCODE plugin of Cytoscape for the purpose of screening of valuable targets. As a result, 915 genes were found to be significantly differential expression in RAS samples and biological processes related to immune and inflammatory response were significantly enriched in those genes. Network and module analysis identified FBXO6, ITGA4, VCAM1 and etc as valuable therapeutic targets for RAS. Finally, FBXO6, ITGA4, and VCAM1 were further confirmed by real time RT-PCR and western blot. This study should be helpful for the research and treatment of RAS.