Makassar — Integrated computational approaches can help accelerate the early stages of natural product-based drug discovery by combining target prediction, network pharmacology, molecular docking, and ADMET prediction. This finding comes from a review by Wahyuni Agus and Fauziah Hasdin of Universitas Negeri Makassar, published in 2026. The review analyzed peer-reviewed scientific literature published between 2015 and 2025 and found that combining these computational methods provides complementary information for identifying biological targets, understanding mechanisms, prioritizing promising compounds, and assessing pharmacokinetic characteristics before experimental validation.
Natural products have long been an important source of bioactive compounds for drug discovery. Their structural diversity and broad biological activities have contributed to the development of therapeutic agents and continue to attract scientific attention as potential sources of new drug candidates. Natural compounds and their derivatives have been explored for applications involving diseases such as cancer, infectious diseases, and inflammatory disorders.
However, developing drugs from natural products can be a lengthy and costly process. Conventional drug discovery may require extensive laboratory screening, target identification, and pharmacological evaluation. Advances in computational biology and bioinformatics have created opportunities to conduct preliminary analyses digitally, allowing researchers to examine potential biological targets, molecular interactions, pharmacokinetic characteristics, and toxicity profiles before proceeding to experimental testing.
Wahyuni Agus and Fauziah Hasdin reviewed four major computational approaches that can be connected into a single workflow. These approaches are target prediction, network pharmacology, molecular docking, and ADMET prediction. Rather than treating them as independent techniques, the review emphasizes their complementary roles in narrowing down promising natural compounds and providing a more comprehensive picture of their potential as drug candidates.
The authors conducted a narrative literature review using scientific publications retrieved from Scopus, Web of Science, PubMed, and Google Scholar. Searches used combinations of keywords related to natural products, computational drug discovery, target prediction, network pharmacology, molecular docking, and ADMET prediction. Peer-reviewed publications from 2015 to 2025 were prioritized and analyzed thematically according to their applications, integration, challenges, and future perspectives.
The integrated workflow begins with natural products, including medicinal plants, marine organisms, microorganisms, and fungi. After compounds are identified, they can undergo a series of computational analyses. The workflow illustrated in Figure 1 on page 4 begins with target prediction, followed by network pharmacology, molecular docking, and ADMET prediction. The combined results are then used to prioritize promising lead compounds before experimental validation through in vitro or in vivo studies.
Target prediction serves as an initial step for identifying potential proteins that may interact with bioactive compounds. According to the review, web-based platforms can predict compound-target interactions using chemical similarity, pharmacophore mapping, and machine learning algorithms. This information provides an early indication of which biological targets may be relevant for further analysis.
The next stage, network pharmacology, provides a broader view of how natural compounds may work through multiple targets and biological pathways. The approach combines compound-target interactions with protein-protein interaction networks and pathway enrichment analysis. This can help researchers identify important therapeutic targets and better understand the mechanisms of natural compounds that may act on multiple biological targets.
Molecular docking is then used to examine interactions between natural compounds and target proteins. The method estimates binding affinity and provides information about how a compound may interact with a particular protein. The review describes molecular docking as a useful tool for supporting target prediction and prioritizing compounds that show potentially favorable interactions.
The fourth stage is ADMET prediction, which covers absorption, distribution, metabolism, excretion, and toxicity. This analysis provides early estimates of pharmacokinetic properties and toxicity profiles. Evaluating these characteristics before experimental studies can help researchers prioritize compounds with more favorable drug-like properties.
The literature reviewed by the authors shows that no single computational method can provide a complete assessment of a natural product’s therapeutic potential. Target prediction helps identify possible protein targets, network pharmacology examines relationships between compounds, targets, and pathways, molecular docking evaluates ligand-protein interactions, and ADMET prediction assesses pharmacokinetic and toxicity characteristics. Combining these approaches therefore produces broader evidence than relying on any single method.
The integrated approach can also make the early stages of drug discovery more efficient. Computational screening allows researchers to narrow down large numbers of compounds before committing resources to laboratory experiments. The review notes that computational drug discovery has increasingly moved from the use of individual in silico techniques toward integrated workflows that can reduce the time and costs associated with conventional screening approaches.
The review also highlights important limitations. Prediction accuracy depends on the quality of chemical and biological databases as well as the algorithms used. Different computational algorithms may produce inconsistent results, while computational models cannot fully reproduce the complexity of biological systems. Experimental validation therefore remains essential for confirming the therapeutic potential of natural compounds.
Future developments may further strengthen computational drug discovery through artificial intelligence, multi-omics data, and molecular dynamics simulations. Machine learning and deep learning have the potential to improve target prediction, virtual screening, and ADMET prediction by analyzing increasingly complex biological datasets. Improvements in public databases and predictive algorithms are also expected to increase the reliability of computational analyses.
For pharmaceutical research and healthcare innovation, the findings offer a framework for more efficient early-stage screening of natural compounds. Researchers can use computational approaches to prioritize candidates from medicinal plants, marine organisms, fungi, microorganisms, and other natural sources before conducting more resource-intensive laboratory studies.
Wahyuni Agus and Fauziah Hasdin conclude that integrating target prediction, network pharmacology, molecular docking, and ADMET prediction provides a practical framework for accelerating natural product-based drug discovery. The combined workflow can support the identification of biological targets, exploration of molecular mechanisms, evaluation of compound-protein interactions, and assessment of pharmacokinetic properties before experimental validation.
The review has several limitations because it focuses on four major computational approaches and does not conduct experimental validation or quantitative meta-analysis. The authors recommend that future studies combine additional computational techniques with experimental validation, particularly through artificial intelligence, molecular dynamics simulations, and multi-omics analyses.
Author Profiles
Wahyuni Agus — Universitas Negeri Makassar.
Fauziah Hasdin — Universitas Negeri Makassar.
Research Source
Article Title: Integrated Computational Approaches for Natural Product-Based Drug Discovery: Target Prediction, Network Pharmacology, Molecular Docking, and ADMET Prediction
Journal: East Asian Journal of Multidisciplinary Research (EAJMR), Vol. 5 No. 8, 2026, pp. 3475–3486.
DOI: https://doi.org/10.55927/eajmr.v5i8.279
Journal Website: https://journaleajmr.my.id/index.php/eajmr
0 Komentar