Phytochemical drug discovery for COVID-19 using high-resolution computational docking and machine learning assisted...

Area of Science:

• Computational chemistry
• Pharmacology
• Virology

Background:

• The COVID-19 pandemic necessitates accessible therapeutics, especially for underserved populations.
• Emerging SARS-CoV-2 variants may resist current vaccines, highlighting the need for alternative treatments.
• Phytochemicals offer potential for polypharmacological approaches against viral infections.

Purpose of the Study:

• To computationally screen a diverse library of phytochemicals for potential SARS-CoV-2 protein inhibition.
• To develop and validate an integrated virtual screening workflow combining structure-based and ligand-based methods.
• To identify novel lead compounds with strong binding affinity and favorable pharmacokinetic properties.

Main Methods:

• Employed structure-based virtual screening (SBVS) to dock phytochemicals against SARS-CoV-2 proteins.
• Utilized ligand-based virtual screening (LBVS) with machine learning to predict additional lead compounds.
• Conducted computational ADME screening to assess pharmacokinetic properties of identified leads.

Main Results:

• Identified 28 phytochemicals exhibiting strong binding interactions with SARS-CoV-2 proteins.
• The integrated LBVS approach increased the lead discovery rate fourfold compared to SBVS alone.
• 18 of the 62 identified leads demonstrated promising pharmacokinetic profiles.

Conclusions:

• The study validates the efficacy of incorporating machine learning into virtual screening for drug discovery.
• This computational workflow accelerates the identification of potential phytochemical therapeutics for COVID-19.
• The identified leads represent promising candidates for further development into accessible antiviral treatments.