2 Network pharmacology-related databases and research tools
With the rapid development of network pharmacology, it has made some breakthroughs in drug development and mechanism research. The concept of network pharmacology has only been born more than ten years, but its value and development potential have been paid more and more attention by scholars. These are inseparable from the improvement of various databases and the breakthrough of research tools, which provide important support for the drug development of network pharmacology. There are many kinds of databases and research tools related to network pharmacology, and different databases and tools have different characteristics and functions. The summary is as follows.
Disease target database and its characteristics: Herb(http://herb.ac.cn/), a special traditional Chinese medicine (TCM) high-throughput experiment and reference database connecting the disease target drug interaction relationship, which also provides a tool for target function analysis; Genecards(https://www.genecards.org/), a database about the detailed information of all human gene proteins at present. When predicting disease targets, as long as the input needs to retrieve the disease, all the corresponding prediction targets will be displayed in the form of scoring, and the operation is relatively simple; Diseases(https://diseases.jensenlab.org/Search) database integrating disease gene association information from the existing database has a large amount of data, but the search time is long; The GEO(https://www.ncbi.nlm.nih.gov/geo/) database is the first public storage database of gene expression data (gene chip, high-throughput sequencing). The arrangement of various diseases in the database is very comprehensive, and the acquisition of disease target data needs to be arranged by yourself; Mir2disease(http://www.mir2disease.org/), a database of human diseases related to miRNA, which provides detailed information about the relationship between miRNA and its diseases; OMIM(https://www.omim.org/), which contains the information database of the association between disease phenotypes and pathogenic genes, and provides the connection, composition, structure, and function of pathogenic genes.
Target prediction database of compounds and its characteristics: Stitch(http://stitch.embl.de/)The database mainly provides protein-protein interactions, which are derived from experiments and literature studies; SEA(https://sea.bkslab.org/)In the database of target prediction based on chemical structure formula, the number of prediction targets is usually small. Targetnet(http://targetnet.scbdd.com/), a database based on 623 human protein designs, which can be used to predict the target of any given compound molecular structure formula; Swisstargetprediction(http://swisstargetprediction.ch/)According to the similarity of the molecular structure formula of the compound, the target of the compound can be predicted, which can be carried out in different species (human, rat, and mouse); Drugbank(https://go.drugbank.com/), one of the most powerful and comprehensive drug databases, with comprehensive information of drug targets. The results have been verified by experiments, but the target acquisition cannot be downloaded in batch; Chembl(https://www.ebi.ac.uk/chembl/)The database contains the therapeutic targets of research drugs and approved drugs, and can also quickly obtain the data related to the biological activity of the target. TTD(http://db.idrblab.net/ttd/)It is a database that provides information related to target genes, including biological pathways, functions, diseases, and drugs corresponding to genes; Pharmmapper(http://lilab.ecust.edu.cn/pharmmapper/check.php)The online small molecule drug target prediction platform based on pharmacophore model has more than 7000 receptor-based pharmacophore models, but the retrieval results are time-consuming.
In addition, network pharmacological algorithms and tools are particularly important for mining these databases. Lei[21] et al proposed an algorithm that can effectively predict the association between drugs and diseases-vgaedr, which is based on variational graph automatic encoder and heterogeneous network. At the same time, algorithms such as DeepDR、SCMFDD、BNNR, and GRGMF are also used to predict the relationship between diseases and drugs. As an open-source tool integrating biomolecular interaction networks and states, Cytoscape can visualize protein interaction networks and annotate data, and modularize regional networks to find core nodes [22]. Autodock Vina is software for molecular docking. It evaluates the binding ability of protein molecules to drug molecules through a specific scoring function, predicts the binding conformation and binding affinity, and the accuracy of receptor-ligand binding mode prediction is also reliable, which will provide a certain value for drug screening and development[23]. In addition, discovery studio, a molecular docking software, is also an analysis tool based on computer simulation to screen potential drugs. Its docking efficiency is better than autodock Vina, which is related to the different docking algorithms used by discovery studio.