Objectives: The aim was to index natural products for less expensive preventive or curative anti-inflammatory therapeutic drugs.
Materials: A set of 441 anti-inflammatory drugs representing the active domain and 2892 natural products representing the inactive domain was used to construct a predictive model for bioactivity-indexing purposes.
Method: The model for indexing the natural products for potential anti-inflammatory activity was constructed using the iterative stochastic elimination algorithm (ISE). ISE is capable of differentiating between active and inactive anti-inflammatory molecules.
Results: By applying the prediction model to a mix set of (active/inactive) substances, we managed to capture 38% of the anti-inflammatory drugs in the top 1% of the screened set of chemicals, yielding enrichment factor of 38. Ten natural products that scored highly as potential anti-inflammatory drug candidates are disclosed. Searching the PubMed revealed that only three molecules (Moupinamide, Capsaicin, and Hypaphorine) out of the ten were tested and reported as anti-inflammatory. The other seven phytochemicals await evaluation for their anti-inflammatory activity in wet lab.
Conclusion: The proposed anti-inflammatory model can be utilized for the virtual screening of large chemical databases and for indexing natural products for potential anti-inflammatory activity.
Keywords: Anti-inflammatory; Bioactivity index; Chemoinformatics; Ligand-based modeling.