Network Pharmacology: Curing Causal Mechanisms Instead of Treating Symptoms

The increasing intricacy of human disorders, especially those influenced by multiple factors, has highlighted the shortcomings of the conventional “one drug–one target” model in pharmacological research. Traditional drug discovery methods often fall short in addressing the complex interplay among numerous genes, proteins, and signaling pathways that drive disease development. This has created a demand for novel approaches that adopt a holistic viewpoint to better decipher disease mechanisms and improve therapeutic interventions.

Network Pharmacology, introduced by Hopkins in 2007, has emerged as a ground-breaking interdisciplinary discipline that combines systems biology, bioinformatics, and pharmacology to study the effects of drugs within complex biological networks [1]. Moving beyond reductionist methods, Network Pharmacology focuses on multi-target therapies and network interactions, offering a more integrated understanding of the relationships between drugs, their targets, and diseases [2,3]. This shift allows for the examination of how drugs influence multiple targets at once, potentially enhancing treatment effectiveness while minimizing side effects.

This field utilizes a blend of computational and experimental techniques to investigate biological systems comprehensively. Computational tools such as network modeling, topological analysis, clustering methods, and data visualization help pinpoint crucial nodes and interactions within biological networks [4]. Meanwhile, experimental approaches—including high-throughput omics technologies like genomics, proteomics, and metabolomics, along with validation through in vitro and in vivo studies—generate critical data to support computational findings [5,6]. The synergy of these methods permits a thorough and systematic investigation of the molecular mechanisms behind drug actions.

In recent years, Network Pharmacology has attracted considerable interest in drug discovery and development. It has been broadly applied to discover new drug candidates, forecast drug–target interactions, and find new therapeutic uses for existing drugs [7,8]. By harnessing extensive biological datasets and advanced computational techniques, Network Pharmacology opens up new target spaces and streamlines the drug development pipeline. Additionally, its capability to integrate diverse data sources, such as genomic, proteomic, and clinical information, makes it a valuable asset in the advancement of precision medicine.[9]

Furthermore, Network Pharmacology plays a crucial role in unravelling the underlying causes of complex disorders. Many conditions result from the disruption of interconnected biological networks rather than a single gene or pathway malfunction. By constructing and analyzing disease-specific networks, researchers can identify key regulatory elements, signaling cascades, and molecular interactions involved in disease progression [2,3]. This comprehensive perspective provides essential insights for designing multi-target therapeutic strategies and identifying biomarkers for diagnosis and prognosis.

Advancements in computational biology, along with the availability of extensive biological databases, have greatly expanded the potential of network pharmacology. The incorporation of artificial intelligence and machine learning into this field has enhanced the precision of predicting drug–target interactions and accelerated the identification of innovative therapeutic approaches. [7,8,10,11]These technological improvements have established network pharmacology as a pivotal method in contemporary biomedical research.

However, despite its many benefits, network pharmacology encounters challenges such as data heterogeneity, inaccuracies within databases, and the necessity for experimental confirmation of computational findings. Overcoming these obstacles is crucial for increasing the robustness and practical utility of network pharmacology. Nonetheless, continuous progress in computational algorithms and experimental methods is reinforcing its promise in drug development and disease investigation.

Overall, network pharmacology marks a transformative shift in pharmacological research by offering a holistic framework to comprehend intricate biological systems and support the creation of safer, more effective treatments. This review seeks to highlight recent progress in network pharmacology, emphasizing its core concepts, techniques, drug discovery applications, and future directions in biomedical science.

2. METHODOLOGY

1. Collection of Natural Compounds and Chemical Space Analysis:
Bioactive natural compounds (NCs) from Clerodendrum species were gathered through an extensive review of scientific literature [2,3]. The two-dimensional (2D) structures of these compounds were obtained from PubChem and ChemSpider databases, compiled into a library, and duplicates were removed. For compounds lacking structural data, 2D structures were manually drawn using Marvin Sketch (v6.2.0) and saved in .mol2 format. These 2D structures were then converted into three-dimensional (3D) conformations using the Corina 3D module within the TSAR platform, and further saved in .pdb format. Energy minimization and hydrogen addition were performed using the CHARMm force field with a smart minimizer algorithm over 1500 steps of steepest descent followed by conjugate gradient optimization, targeting a convergence gradient of 0.001 kcal/mol. Additionally, 43 approved anticancer drugs were retrieved from Drug Bank as reference compounds.[12] Molecular descriptors for both NCs and reference drugs were calculated using PaDEL-Descriptor. Principal Component Analysis (PCA) was conducted with Bio Vinci software to assess compound distribution within chemical space, utilizing eight molecular descriptors such as a LogP, molecular weight, hydrogen bond donors and acceptors, rotatable bonds, ring count, aromatic rings, and fractional polar surface area.

2. ADME and Toxicity Profiling:
Pharmacokinetic properties including absorption, distribution, metabolism, excretion, and toxicity (ADME-T) of the selected compounds were predicted using the PreADMET server and Osiris Property Explorer. These analyses helped evaluate the drug-likeness and safety profiles of the compounds.[7,10]

3. Retrieval and Preparation of Target Proteins:
Protein targets relevant to the study were sourced from multiple databases including the Therapeutic Target Database, Drug Bank, Uniport, and the Protein Data Bank.[12,13] Redundant or incomplete protein structures and those lacking active binding sites were excluded. Only human (Homo sapiens) proteins were considered. A total of 60 high-quality protein structures, determined by X-ray crystallography with resolutions between 1.2 Å and 3.5 Å, were selected and processed using UCSF Chimera to prepare and optimize the structures.

4. Protein–Protein Interaction (PPI) Network Analysis:
Protein interaction data were obtained from STRING (version 11.0)[14] with a focus on Homo sapiens and filtered to include interactions with a confidence score above 0.9 to ensure reliability. Disconnected nodes were removed. The resulting network data were visualized and analysed using Cytoscape (v3.7.2),[4] enabling construction of target–pathway networks to identify critical proteins and signaling pathways involved in disease progression.

5. Gene Ontology (GO) and KEGG Pathway Enrichment Analysis:
Functional enrichment analyses, including Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway mapping, were conducted via the DAVID database.[15] Official gene symbols for Homo sapiens were used to identify significant biological processes, molecular functions, and pathways associated with the target proteins.[16]

6. Molecular Docking Analysis:
Molecular docking was performed using PyRx with the AutoDock Vina algorithm.[17] Blind docking was employed to explore potential binding sites across entire protein surfaces. Docking outcomes were assessed based on binding affinity scores, and top-ranked compound-protein interactions underwent detailed analysis to examine binding modes and stability.

7. Network Pharmacology Analysis:
Network pharmacology approaches were applied to assess the poly pharmacological potential of selected compounds. Three types of networks were constructed and analysed using Cytoscape with the Network Analyser plugin: compound–target, target–pathway, and compound–target–pathway networks.[4]

8. Molecular Dynamics Simulation:
Molecular dynamics (MD) simulations were conducted with GROMACS 2021 on a Linux Ubuntu platform. Topology files were generated using the CHARMM36 force field and Swiss PARAM server. Systems were solvated with the TIP3P water model and neutralized with Na?/Cl? ions. Energy minimization was performed via the steepest descent algorithm (50,000 steps), followed by equilibration under NVT and NPT ensembles at 300 K for 100 ps each. A 10 ns production run was carried out with a 2 fs time step. Trajectory analyses including root mean square deviation (RMSD), root mean square fluctuation (RMSF), radius of gyration (Rg), and hydrogen bond analysis were performed to evaluate the stability of protein–ligand complexes.

3. DISTRIBUTION OF DATABASE CATEGORIES UTILIZED

The reviewed studies employed a broad range of computational databases and tools for compound identification, target screening, and mechanistic validation, grouped into six major categories:

3.1 Herbal and Phytochemical Databases:

· IMPPAT (Indian Medicinal Plants, Phytochemistry And Therapeutics): Covers Indian medicinal plants and their phytochemicals with therapeutic associations (17.6%).

· TCM Database@Taiwan: Contains extensive TCM prescriptions and compound data (11.8%).

· HERB: Integrates herbal compounds with transcriptomic and proteomic data (5.9%).

· ETCM (Encyclopedia of Traditional Chinese Medicine): Provides herbal formulas, chemical constituents, and target predictions (5.9%).

· NPASS (Natural Product Activity and Species Source Database): Contains experimentally validated bioactivities linked to species (5.9%).

These repositories are instrumental in sourcing bioactive natural compounds and linking them to therapeutic applications [18,19].

3.2 Chemical Structure and ADMET Databases:

· PubChem: The most widely used (31.25%), a comprehensive repository of chemical structures and bioactivity data.

· SwissADME: Used to predict drug-likeness and ADME properties (18.75%).

· ChEMBL, admetSAR, ChemDraw, NIST Chemistry WebBook, Dictionary of Natural Products, and CDRUG: These databases provide bioactivity data, ADMET predictions, chemical drawing and verification tools, thermochemical data, and focus on anticancer compounds respectively, each contributing around 6.25% usage.

These resources support compound identification, structural validation, and pharmacokinetic profiling [7,8,10].

3.3 Target Prediction Platforms:

· Swiss Target Prediction: The most applied platform (36.36%) using chemical similarity to predict protein targets.

· Other platforms include DIGEP-Pred, SEA, Pharm Mapper, Way2Drug, SuperPred, ImaGEO, STITCH, STRING, TTD, and PharmGKB, each contributing 4.55% to 9.09% usage. These tools facilitate prediction of molecular targets, drug-target interactions, and functional genomics integration [14,7].

3.4 Disease-Gene Annotation Databases:

· Gene Cards: Most frequently used (44.44%), linking genes to functions, pathways, and diseases.

· OMIM and DisGeNET: Both at 22.22%, provide clinically annotated genetic disorder data and curated disease-gene associations.

· Other databases such as GEO, NCBI Gene, and HPO also contribute, connecting genetic information to disease phenotypes [20,21].

3.5 Protein Annotation and Interaction Databases:

· Uniport/UniProtKB: Used in 62.5% of cases for protein sequence, structure, and functional annotation.

· BindingDB and STITCH: Used to validate protein-ligand interactions and chemical-protein networks, contributing 25% and 12.5% usage respectively [13,].

3.6 Multi-Database Integrators:
Platforms such as TCMIP v2.0, HERB-linked compound frameworks, Metascape, and hybrid approaches combining metabolomics data with disease-gene repositories (e.g., Drug Bank, OMIM, DisGeNET, Genecards) enhance compound annotation, target prediction reliability, and holistic network construction. These integrators support the cross-validation and integration of diverse biological data [12,22].

4. NETWORK PHARMACOLOGY-SPECIFIC PLATFORMS

These specialized computational platforms are designed to integrate and analyse large-scale biological and pharmacological data sets, enabling the exploration of drug actions within complex disease networks. They emphasize multi-target effects and systems-level understanding of therapeutic mechanisms:

1. TCMSP (Traditional Chinese Medicine Systems Pharmacology Database): The most frequently cited (52.9%), offering herbal ingredients, ADMET properties, and drug-target information for systems pharmacology and network pharmacology applications. A comprehensive database integrating traditional Chinese medicine components with pharmacokinetic and pharmacodynamics data. It supports drug-target prediction and molecular interaction network construction, widely used for identifying multi-target drugs within TCM [23,24].

2. STRING: Provides extensive data on known and predicted protein-protein interactions, integrating experimental data, text mining, and computational predictions. It is commonly used to build interaction networks and uncover molecular mechanisms related to diseases and drug targets [14].

3. KEGG: A bioinformatics resource offering detailed information on biological pathways, diseases, and drug interactions. KEGG pathway maps facilitate system-level analysis of drug effects on metabolic and signaling pathways [16,25].

4. Cytoscape: An open-source platform for visualizing and analyzing molecular interaction networks. It supports integration of diverse biological data, network topology analysis, and pathway enrichment, making it essential for drug-target network visualization and repurposing studies [4].

5. Drug Bank: An extensive database detailing FDA-approved and experimental drugs, including mechanisms of action, side effects, and interactions with genes and proteins. It is extensively used for drug discovery and target prediction [12,24].

6. Open Targets Platform: Integrates genetic, functional genomics, and chemical data to elucidate drug-disease relationships, aiding in the identification of drug targets and biomarkers through data-driven approaches [26].

7. PharmGKB: Curates pharmacogenomics data regarding genetic variation impacts on drug response, supporting personalized medicine by linking genetic variants to drug metabolism and efficacy [9].

5. Molecular Docking and Virtual Screening Tools

Molecular docking and virtual screening constitute fundamental computational approaches in modern drug discovery, enabling the prediction of interactions between small molecule ligands and biological macromolecules such as proteins or nucleic acids. [27]These techniques play a critical role in identifying potential drug candidates, elucidating binding affinities, and understanding structure-activity relationships, thereby accelerating the drug development process.

Molecular docking involves the computational simulation of a ligand’s optimal binding pose within the active site of a target molecule. This process typically includes generating multiple ligand conformations (pose prediction) and evaluating their binding affinities using scoring functions that quantify molecular interactions, including hydrogen bonding, hydrophobic contacts, and van der Waals forces. Importantly, docking protocols often implement flexible docking strategies that account for conformational changes in both the ligand and receptor to more accurately mimic physiological conditions.[17,27}

Virtual screening extends this concept by computationally evaluating large libraries of chemical compounds to identify molecules with high likelihood of binding to a specific biological target. Structure-based virtual screening (SBVS) employs the three-dimensional structure of the target protein to guide ligand selection, whereas ligand-based virtual screening (LBVS) relies on the chemical and structural features of known active compounds to identify analogs with similar properties.

A variety of software tools support these methodologies. AutoDock and AutoDock Vina are widely used open-source programs known for their efficiency and ease of integration with visualization platforms like PyMOL. Commercial suites such as Schrödinger’s Glide offer high-precision docking with advanced scoring algorithms and receptor flexibility. The Molecular Operating Environment (MOE) provides an integrated environment for docking, virtual screening, and pharmacophore modeling. Swiss Dock offers a web-based docking solution utilizing CHARMM force fields, while GOLD specializes in genetic algorithm-based flexible docking with robust scoring functions. Cloud-based tools like DockThor facilitate flexible and accurate docking studies with scalable resources. [17,28,29]Complementary virtual screening tools include the ZINC database, which provides access to extensive libraries of commercially available compounds, and Ligand Scout, which generates pharmacophore models for ligand-based screening.[30]

In recent years, artificial intelligence (AI) and machine learning (ML) techniques have been incorporated into virtual screening workflows. Platforms such as Deep Dock and Deep Screening utilize deep learning algorithms to improve the accuracy of docking predictions and compound ranking. These technological advances have significantly expanded the scope and efficiency of computational drug discovery.[7,8,10,11]

Applications of molecular docking and virtual screening are diverse, encompassing lead compound identification, drug repurposing for novel therapeutic indications, and detailed investigation of molecular mechanisms underlying drug-target interactions. These approaches significantly reduce the time and cost associated with experimental screening, making them indispensable tools in contemporary pharmacological research.

6. Cross-Platform Integration in Network Pharmacology

Cross-platform integration is a pivotal strategy in network pharmacology that involves the combination of diverse computational tools, databases, and multi-omics datasets to construct unified models of biological systems. This integrative approach enhances the understanding of complex disease mechanisms and facilitates the identification of novel therapeutic targets and strategies.[5,6]

One key aspect of cross-platform integration is the incorporation of multi-omics data. Large-scale projects such as The Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression (GTEx) project generate comprehensive datasets spanning genomics, transcriptomic, proteomics, and metabolomics. Integrating these data types with network pharmacology models enables the construction of highly detailed, disease-specific molecular networks. This holistic view allows researchers to identify critical nodes and pathways that may serve as drug targets or biomarkers.[14,4,16]

Software and database integration also plays a crucial role. For instance, Cytoscape, a powerful network visualization tool, can be combined with protein interaction databases like STRING and pathway resources such as KEGG to construct and analyse complex target-pathway networks. [31]Enrichment analysis platforms like Metascape and DAVID further refine these networks by identifying significantly associated biological processes and pathways, thereby enhancing the biological relevance of network pharmacology findings.

Interoperability among various computational environments (e.g., Python, R, Cytoscape) and specialized software tools such as RDKit for chemoinformatics and PyMOL for molecular visualization allows for seamless data flow and multi-dimensional analysis. This flexibility enables researchers to integrate chemical, genomic, and structural data within a unified analytical framework.[7,8,10]

The integration of AI and machine learning technologies into network pharmacology is another emerging frontier. Algorithms implemented in platforms like DeepChem and BioBERT facilitate automated drug candidate identification and predictive modeling of drug efficacy. By cross-validating results across multiple platforms, these AI-driven approaches improve the reliability and robustness of network pharmacology analyses.

7. Future Trends and Emerging Tools in Network Pharmacology

The future of network pharmacology is shaped by rapid advances in computational power, AI, and the integration of multi-omics data, all of which are revolutionizing drug discovery, disease modeling, and precision medicine.

Artificial intelligence and machine learning are increasingly utilized to analyse vast biomedical datasets, predict drug-target interactions with higher accuracy, and uncover novel therapeutic targets. Techniques such as deep learning and reinforcement learning enable the modeling of complex biological systems and optimization of drug discovery pipelines. Tools like DeepChem and chemoinformatics platforms harness AI to predict drug efficacy, toxicity, and polypharmacology, thereby reducing reliance on costly and time-consuming experimental methods.[7,8,10]

Multi-omics integration remains a cornerstone of future developments, with new computational tools designed to synthesize genomic, proteomic, transcriptomic, and metabolomics data layers. Platforms such as Open Targets and BioGRID facilitate the incorporation of large-scale omics datasets into network pharmacology workflows, providing a more comprehensive understanding of disease biology and facilitating the discovery of multi-target therapeutic strategies.[5,6,26]

Personalized and precision medicine represents another promising avenue, wherein individual patient data—including genetic profiles and clinical histories—are integrated with computational models to tailor treatments for maximum efficacy and minimal side effects. Resources such as PharmGKB and PRISM are instrumental in linking genetic variation to drug response, guiding personalized therapy decisions.[9,32]

AI-driven drug repurposing is gaining traction as a cost-effective strategy for identifying new uses for existing drugs. Databases like RepurposeDB and the Drug Repurposing Hub leverage network analysis combined with machine learning to accelerate this process, offering rapid solutions for emerging or neglected diseases.

Advancements in network analysis platforms, including Gephi, Cytoscape, and Ingenuity Pathway Analysis (IPA), continue to enhance capabilities for network construction, visualization, and interpretation. The integration of graph-based AI techniques further improves detection of hidden patterns within complex biological datasets.

Nanotechnology coupled with network pharmacology is emerging as a powerful approach for designing targeted drug delivery systems. Tools like Nano HUB enable modeling of nanoparticle interactions with biological systems, fostering the development of smart nanodrugs that improve therapeutic efficacy and reduce toxicity.[33]

Finally, cloud computing and big data analytics are becoming indispensable for managing the exponentially growing volume of biomedical data. Platforms such as Amazon Web Services (AWS), Google Cloud, and Microsoft Azure offer scalable infrastructure for data storage and computational analysis, facilitating the identification of intricate disease networks and accelerating personalized medicine development.

CONCLUSION

Network pharmacology represents a transformative paradigm in drug discovery, bridging traditional medicinal knowledge with modern computational approaches to address complex biomedical challenges. By integrating bioactive natural compounds with their molecular targets, this systems-level approach elucidates the mechanisms of action of traditional formulations and supports the discovery and repurposing of therapeutic agents.

While current network pharmacology analyses depend largely on existing datasets, which are continually evolving with ongoing research, the approach remains invaluable for uncovering novel insights into disease biology and drug action. Despite inherent limitations such as data heterogeneity and the need for experimental validation, network pharmacology provides a robust framework for analyzing complex biological information.

Ultimately, this integrative methodology advances our understanding of multifaceted disease mechanisms, enabling the development of safer, more effective, and personalized therapeutic strategies. It holds significant promise for overcoming current limitations in drug discovery and delivering innovative solutions to pressing medical challenges.

REFERENCES

1. Hopkins AL. Network pharmacology: the next paradigm in drug discovery. Nature chemical biology. 2008 Nov;4(11):682-90.
2. Barabási AL, Gulbahce N, Loscalzo J. Network medicine: a network-based approach to human disease. Nature reviews genetics. 2011 Jan;12(1):56-68.
3. Berger SI, Iyengar R. Network analyses in systems pharmacology. Bioinformatics. 2009 Oct 1;25(19):2466-72.
4. Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, Amin N, Schwikowski B, Ideker T. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome research. 2003 Nov 1;13(11):2498-504.
5. Argelaguet R, Velten B, Arnol D, Dietrich S, Zenz T, Marioni JC, Buettner F, Huber W, Stegle O. Multi-Omics Factor Analysis—a framework for unsupervised integration of multi-omics data sets. Molecular systems biology. 2018 Jun 1;14(6):MSB178124.
6. Chong J, Soufan O, Li C, Caraus I, Li S, Bourque G, Wishart DS, Xia J. MetaboAnalyst 4.0: towards more transparent and integrative metabolomics analysis. Nucleic acids research. 2018 Jul 2;46(W1):W486-94.
7. Jiménez-Luna J, Grisoni F, Weskamp N, Schneider G. Artificial intelligence in drug discovery: recent advances and future perspectives. Expert opinion on drug discovery. 2021 Sep 2;16(9):949-59.
8. Vamathevan J, Clark D, Czodrowski P, Dunham I, Ferran E, Lee G, Li B, Madabhushi A, Shah P, Spitzer M, Zhao S. Applications of machine learning in drug discovery and development. Nature reviews Drug discovery. 2019 Jun;18(6):463-77.
9. Raj GM. Pharmacogenetics, pharmacogenomics, and personalized medicine. InIntroduction to Basics of Pharmacology and Toxicology: Volume 1: General and Molecular Pharmacology: Principles of Drug Action 2019 Nov 17 (pp. 235-259). Singapore: Springer Singapore.
10. Chen S, Li T, Yang L, Zhai F, Jiang X, Xiang R, Ling G. Artificial intelligence-driven prediction of multiple drug interactions. Briefings in bioinformatics. 2022 Nov;23(6):bbac427.
11. Frusciante L, Visibelli A, Geminiani M, Santucci A, Spiga O. Artificial intelligence approaches in drug discovery: towards the laboratory of the future. Current Topics in Medicinal Chemistry. 2022 Oct 1;22(26):2176-89.
12. Wishart DS, Feunang YD, Guo AC, Lo EJ, Marcu A, Grant JR, Sajed T, Johnson D, Li C, Sayeeda Z, Assempour N. DrugBank 5.0: a major update to the DrugBank database for 2018. Nucleic acids research. 2018 Jan 4;46(D1):D1074-82.
13. UniProt Consortium. UniProt: a worldwide hub of protein knowledge. Nucleic acids research. 2019 Jan 8;47(D1):D506-15.
14. Szklarczyk D, Gable AL, Nastou KC, Lyon D, Kirsch R, Pyysalo S, Doncheva NT,
15. Legeay M, Fang T, Bork P, Jensen LJ. The STRING database in 2021: customizable protein–protein networks, and functional characterization of user-uploaded gene/measurement sets. Nucleic acids research. 2021 Jan 8;49(D1):D605-12.
16. Huang DW, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nature protocols. 2009 Jan;4(1):44-57.
17. Kanehisa M, Goto S. KEGG: kyoto encyclopedia of genes and genomes. Nucleic acids research. 2000 Jan 1;28(1):27-30.
18. Trott O, Olson AJ. AutoDock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. Journal of computational chemistry. 2010 Jan 30;31(2):455-61.
19. Shao LI, Zhang B. Traditional Chinese medicine network pharmacology: theory, methodology and application. Chinese journal of natural medicines. 2013 Mar 1;11(2):110-20.
20. Cao S, Han Y, Li Q, Chen Y, Zhu D, Su Z, Guo H. Mapping pharmacological network of multi-targeting litchi ingredients in cancer therapeutics. Frontiers in Pharmacology. 2020 Apr 24;11:451.
21. Piñero J, Bravo À, Queralt-Rosinach N, Gutiérrez-Sacristán A, Deu-Pons J, Centeno E, García-García J, Sanz F, Furlong LI. DisGeNET: a comprehensive platform integrating information on human disease-associated genes and variants. Nucleic acids research. 2016 Oct 19:gkw943.
22. Wang Y, Li Y, Chen J, Lai L. Modeling protein–ligand interactions for drug discovery in      the era of deep learning. Chemical Society Reviews. 2025.
23. Zhou Y, Zhou B, Pache L, Chang M, Khodabakhshi AH, Tanaseichuk O, Benner C, Chanda SK. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nature communications. 2019 Apr 3;10(1):1523.
24. Ru, J., et al. (2014). TCMSP: A database of systems pharmacology for drug discovery from herbal medicines. Journal of Cheminformatics , 6, 13.
25. Ru J, Li P, Wang J, Zhou W, Li B, Huang C, Li P, Guo Z, Tao W, Yang Y, Xu X. TCMSP: a database of systems pharmacology for drug discovery from herbal medicines. Journal of cheminformatics. 2014 Apr 16;6(1):13
26. Kanehisa M, Furumichi M, Tanabe M, Sato Y, Morishima K. KEGG: new perspectives on genomes, pathways, diseases and drugs. Nucleic acids research. 2017 Jan 4;45(D1):D353-61.
27. Koscielny G, An P, Carvalho-Silva D, Cham JA, Fumis L, Gasparyan R, Hasan S, Karamanis N, Maguire M, Papa E, Pierleoni A. Open Targets: a platform for therapeutic target identification and validation. Nucleic acids research. 2017 Jan 4;45(D1):D985-94.
28. Kitchen DB, Decornez H, Furr JR, Bajorath J. Docking and scoring in virtual screening for drug discovery: methods and applications. Nature reviews Drug discovery. 2004 Nov;3(11):935-49.
29. Friesner RA, Banks JL, Murphy RB, Halgren TA, Klicic JJ, Mainz DT, Repasky MP, Knoll EH, Shelley M, Perry JK, Shaw DE. Glide: a new approach for rapid, accurate docking and scoring. 1. Method and assessment of docking accuracy. Journal of medicinal chemistry. 2004 Mar 25;47(7):1739-49.
30. Grosdidier A, Zoete V, Michielin O. SwissDock, a protein-small molecule docking web service based on EADock DSS. Nucleic acids research. 2011 May 28;39(suppl_2):W270-7.
31. Irwin JJ, Shoichet BK. ZINC− a free database of commercially available compounds for virtual screening. Journal of chemical information and modeling. 2005 Jan 24;45(1):177-82.
32. Team RC. RA language and environment for statistical computing, R Foundation for Statistical. Computing. 2020.
33. Sirota M, Dudley JT, Kim J, Chiang AP, Morgan AA, Sweet-Cordero A, Sage J, Butte AJ. Discovery and preclinical validation of drug indications using compendia of public gene expression data. Science translational medicine. 2011 Aug 17;3(96):96ra77-.
34. Li J, Long Q, Ding H, Wang Y, Luo D, Li Z, Zhang W. Progress in the treatment of central nervous system diseases based on nanosized traditional Chinese medicine. Advanced Science. 2024 Apr;11(16):2308677.