Computational Drug Discovery of Natural Product Compounds for Tyrosinase Inhibition

Plants produce chemically diverse compounds for defence and survival, many of which exhibit biological activity and offer promising starting points for drug discovery. This study thus investigated natural compounds as potential inhibitors of tyrosinase, a key enzyme in melanin production and an established target in the treatment of hyperpigmentation. A quantitative structure–activity relationship (QSAR) machine-learning model was developed using ChEMBL tyrosinase bioactivity data and property-matched decoys. Ten models, including decision trees, random forests and gradient boosting, were evaluated; logistic regression was selected for virtual screening. To make the model generalise to unseen chemistry, the model was tested on compounds that were required to have Tanimoto similarity of 0.40 or less to every training compound and to share no Bemis–Murcko scaffold with them. The model achieved an AUC of 0.82 (95% confidence interval [CI], 0.78–0.86), compared with 0.56 using physicochemical properties alone. The frozen model was then used to screen 404,570 natural products from COCONUT (COlleCtion of Open Natural prodUcTs) and IMPPAT (Indian Medicinal Plants, Phytochemistry And Therapeutics). Of the top 1% of predictions, 1,694 compounds were both structurally novel and represented new scaffolds. Following computational filters for predicted skin permeability and potentially reactive chemical groups indicating skin irritation, the 100 highest-scoring candidates were selected for molecular docking. Docking reproduced the crystallographic tropolone binding pose with a root-mean-square deviation (RMSD) of 1.20 Å, while the most predictive docking score achieved an AUC of 0.64 (95% CI, 0.61–0.66). Among the ten highest-ranked candidates, two were associated with Anisomeles indica (Indian catmint) and two with Phaseolus lunatus (lima beans). These candidates warrant experimental validation and further investigation for potential dermatological applications.

Authors

Institutions

Publication Details

Journal
ChemRxiv
Published
2026-11-16
DOI
https://doi.org/10.26434/chemrxiv.15012270/v1
Primary Topic
Computational Drug Discovery Methods
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Computational Drug Discovery of Natural Product Compounds for Tyrosinase Inhibition

Anand Sharma
3 links
ChemRxiv
Computational Drug Discovery Methods
preprint

Computational Drug Discovery of Natural Product Compounds for Tyrosinase Inhibition

Anand Sharma
preprint en

Abstract

Plants produce chemically diverse compounds for defence and survival, many of which exhibit biological activity and offer promising starting points for drug discovery. This study thus investigated natural compounds as potential inhibitors of tyrosinase, a key enzyme in melanin production and an established target in the treatment of hyperpigmentation. A quantitative structure–activity relationship (QSAR) machine-learning model was developed using ChEMBL tyrosinase bioactivity data and property-matched decoys. Ten models, including decision trees, random forests and gradient boosting, were evaluated; logistic regression was selected for virtual screening. To make the model generalise to unseen chemistry, the model was tested on compounds that were required to have Tanimoto similarity of 0.40 or less to every training compound and to share no Bemis–Murcko scaffold with them. The model achieved an AUC of 0.82 (95% confidence interval [CI], 0.78–0.86), compared with 0.56 using physicochemical properties alone. The frozen model was then used to screen 404,570 natural products from COCONUT (COlleCtion of Open Natural prodUcTs) and IMPPAT (Indian Medicinal Plants, Phytochemistry And Therapeutics). Of the top 1% of predictions, 1,694 compounds were both structurally novel and represented new scaffolds. Following computational filters for predicted skin permeability and potentially reactive chemical groups indicating skin irritation, the 100 highest-scoring candidates were selected for molecular docking. Docking reproduced the crystallographic tropolone binding pose with a root-mean-square deviation (RMSD) of 1.20 Å, while the most predictive docking score achieved an AUC of 0.64 (95% CI, 0.61–0.66). Among the ten highest-ranked candidates, two were associated with Anisomeles indica (Indian catmint) and two with Phaseolus lunatus (lima beans). These candidates warrant experimental validation and further investigation for potential dermatological applications.

ChemRxiv
Indian Institute of Technology Kanpur (IN)
Good health and well-being
Computational Drug Discovery Methods
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon