Preliminary data were collected from the PubMed database. Retrieval keywords were subjected with “calycosin,” “heart failure, and “ “diabetes mellitus” (MeSH) terms. In addition, literatures were restricted to the journal original paper and English language. The databases Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP), Chemical-Protein Interaction Networks (STITCH; http://stitch.embl.de/), Swiss Target Prediction, SuperPred webserver, Bioinformatics Analysis Tool for Molecular mechANism of Traditional Chinese Medicine (BATMAN-TCM; http://bionet.ncpsb.org/batman-tcm/), ChemMapper, and Drug Target Prediction System were collectively used to screen the functional genes/targets of calycosin. Furthermore, the HF- and DM-driven genes/targets were acquired from the detection databases DisGeNET, GeneCard, DrugBank (www.drugbank.ca), PharmDB-K, and Therapeutic Target Database (TTD). After mapping the genes/targets of calycosin and HF/DM using an online website (http://bioinformatics.psb.ugent.be/webtools/Venn/), the predictive targets of calycosin against HF were determined (18, 19).

In brief, all mapped predictive targets were further analyzed after creating a target–target-associated functional interprotein network and protein–protein interaction (PPI) network using TSV data from the STRING database. The final hub targets of calycosin against HF/DM were obtained using the Network Analyzer tool in Cytoscape (v3.7.1) software, calculating from the degree values of topology parameters (20, 21).

The molecular structure of the niacin compound was obtained from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/). The protein structures of HF/DM targets were obtained from the Protein Data Bank database (https://www.rcsb.org/). Molecular mechanics-2 force field optimization was conducted using the three-dimensional structure of the compound downloaded using the ChemBio3D Draw module in the ChemBioOffice software (version 2010). The pdbqt structure file necessary for virtual screening was created using the Raccoon software. Next, the docking active center (including all residues around the original ligand) was set using the grid box function in the software, according to the root mean square deviation (RMSD) between the docked ligand and original ligand molecules with appropriate docking parameters. RMSD ≤4 is generally believed to be the threshold for the ligand conformation after docking to match the original ligand conformation (22, 23).

Briefly, the ClusterProfiler, ReactomePA, and AnnotationHub packages in R software (3.6.1) were used to perform a gene ontology (GO)–based biological process and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses of hub targets. Finally, the bubble charts, graphs, and histograms were obtained and visualized (24, 25).

Furthermore, using the Cytoscape software (https://cytoscape.org/), the data of hub targets in the biological processes and pathways of calycosin against HF/DM were used to construct drug–target–GO and function–pathway–disease visualization graphics (26, 27).

Following the bioinformatics analysis, the accessible data revealed 1,499 and 2,359 HF- and DM-related genes, respectively. Furthermore, 113 genes of calycosin were found after screening and removing duplicates using the UniProt database (https://www.uniprot.org/). As shown in Figure 1 , 39 mapped targets are displayed in a Venn diagram. Furthermore, these PPI data with 39 targets were collected for constructing a target–target functional interaction network.

The mapped intersection proteins were imported into Cytoscape software to determine the topological parameters of calycosin against HF/DM. Screening parameters included betweenness, closeness, and connectivity, and the yellow node represented the final screening result. Twenty hub targets highly correlated with calycosin against HF/DM were obtained. These included mitogen-activated protein kinase-1 (MAPK1), ARRB1, ABL1, CDK1, MAP3K3, STUB1, PPP1CA, STAT3, PML, IQGAP1, HSPA9, CAV1, KPNB1, SQSTM1, RUVBL1, SFPQ, PPP2CA, PCBP1, ANXA2, and PAK1 ( Figure 2 ).

The enrichment analyses showed that the main GO-based biological processes of hub targets were involved in the positive regulation of protein phosphorylation, apoptotic process, regulation of cell adhesion, regulation of circadian rhythm, proteasome-mediated ubiquitin-dependent protein catabolic process, positive regulation of fibroblast proliferation, negative regulation of protein binding, positive regulation of protein ubiquitination, positive regulation of peptidyl-serine phosphorylation, MAPK cascade, cell–cell adhesion, caveolin-mediated endocytosis, activation of MAPK activity, Fc-gamma receptor signaling pathway involved in phagocytosis, Bergmann glial cell differentiation, negative regulation of the apoptotic process, response to hypoxia, protein autophosphorylation, Janus Kinase (JAK)-Signal Transducer and Activator of Transcription (STAT) cascade involved in the growth hormone signaling pathway, and positive regulation of dendrite development. The cellular components (CCs) of hub targets were related to cytosol, cytoplasm, nucleoplasm, extracellular exosome, ruffle, focal adhesion, cell–cell junction, nuclear membrane, cell–cell adherens junction, nuclear matrix, midbody, nuclear chromosome, telomeric region, protein complex, microtubule cytoskeleton, nucleus, basolateral plasma membrane, pseudopodium, membrane raft, membrane, and axon. The molecular functions (MFs) of the hub targets included protein binding, ubiquitin–protein ligase binding, protein kinase binding, cadherin binding involved in cell–cell adhesion, ATP binding, protein kinase activity, protein serine/threonine kinase activity, S100 protein binding, RNA polymerase II carboxy-terminal domain kinase activity, mitogen-activated protein kinase binding, transcription regulatory region DNA binding, Hsp90 protein binding, poly(A) RNA binding, Hsp70 protein binding, transcription factor binding, Rac Guanosine Triphosphatase (GTPase) binding, Smad binding, phosphoprotein phosphatase activity, identical protein binding, and protein kinase C binding ( Figures 3A–C ). In addition, 16 KEGG pathways involving hub targets were determined (P < 0.05). The primary signaling pathways comprised proteoglycans in cancer, oocyte meiosis, acute myeloid leukemia, the chemokine signaling pathway, focal adhesion, regulation of actin cytoskeleton, Epidermal Growth Factor Receptor (ErbB) signaling pathway, MAPK signaling pathway, neurotrophin signaling pathway, axon guidance, hepatitis C, adrenergic signaling in cardiomyocytes, pathways in cancer, herpes simplex infection, Cyclic Adenosine Monophosphate (cAMP) signaling pathway, and viral carcinogenesis ( Figures 4A–C ). The network visualization of calycosin–target–GO–KEGG–HF/DM was achieved using Cytoscape software ( Figure 5 ).

As shown in Figure 6A , the active cavity box model of MAPK1 has parameter settings of center-x-y-z as 6.89, -3.188, and 16.85 and size-x-y-z as 15, 15, and 15, and the RMSD of the original ligand is 2.73 Å. The hydrogen bonding of the pro-ligand FRZ to the 1Tvo protein involved the amino acid residues ASP-106 (3.0Å), LYS-54 (3.5Å), GLN-105 (3.3Å), and CYS-166 (2.5 Å) and the hydrogen bond formation with calycosin included SER-153 (1.8 Å), ASN-154 (2.8Å), LYS-54 (3.0Å), ILE-103 (2.3Å), GLN-105 (3.3Å), and CYS-166 (2.5 Å) ( Figure 6B ).

As shown in Figure 6C , the active cavity box model of ARRB1 has center-x-y-z as -22.343, 57.812, and 3.635 and size-x-y-z as 14, 14, and 14, and the RMSD of the original ligand is 2.18 Å. Hydrogen bonding was formed between the pro-ligand IHP and 1ZSH protein through the amino acid residues LYS-326 (2.4Å), LYS-232 (3.2Å), LYS-324 (3.5Å), ARG-236 (2.0 Å), and LYS-250 (2.1 Å), and calycosin formed hydrogen bonds with LYS-326 (2.4Å), LYS-232 (3.2Å), and LYS-324 (3.5Å) ( Figure 6D ).

As shown in Figure 6E , the active cavity box model of ABL1 has center-x-y-z as 13.039, 95.919, and 58.059 and size-x-y-z as 30, 30, and 30, and the RMSD of the original ligand is 1.97Å. The hydrogen bonding between the pro-ligand Polycomb Repressive Complex (PRC) and 1FPU protein involved the amino acid residues MET-318 (1.9 Å), VAL-379 (3.1 Å), and GLU-286 (3.4 Å), and calycosin formed hydrogen bonds with MET-318 (2.4 Å) and GLY-321 (2.7Å) ( Figure 6F ).