A novel approach to active compounds identification based on support vector regression model and mean impact value

Area of Science:

• Natural Product Chemistry
• Computational Chemistry
• Pharmacology

Background:

• Traditional natural product drug discovery is costly and inefficient.
• Bioassay-guided fractionation is a common but resource-intensive method.
• Identifying active compounds from complex mixtures like curcuminoids presents challenges.

Purpose of the Study:

• To develop and validate a computational method for identifying lead active compounds from Curcuma longa L. curcuminoids.
• To predict the cytotoxic potential of curcuminoids against HeLa cells.
• To offer an efficient and economical alternative to traditional drug discovery methods.

Main Methods:

• Support vector regression model trained on mean impact value.
• Identification of high-impact curcuminoid constituents.
• Cytotoxicity assessment using MTT assays and comparison with literature data.

Main Results:

• Eight curcuminoid constituents were identified as having significant cytotoxicity.
• The 50% inhibiting concentrations (IC50) for curcumin, demethoxycurcumin, and bisdemethoxycurcumin were determined.
• The computational method successfully predicted lead active compounds prior to experimental separation.

Conclusions:

• Support vector regression combined with mean impact value analysis is an efficient and economical approach for natural product drug discovery.
• This method can significantly accelerate the identification of potential anti-cancer agents.
• The findings pave the way for faster development of new therapeutics from natural sources.