Machine-Learning-Enabled Virtual Screening for Inhibitors of Lysine-Specific Histone Demethylase 1.

Area of Science:

• Computational chemistry
• Medicinal chemistry
• Machine learning

Background:

• Lysine specific demethylase 1 (LSD1) is a key target in cancer therapy.
• Developing novel LSD1 inhibitors is crucial for anti-cancer drug discovery.

Purpose of the Study:

• To apply machine learning for virtual screening of lysine specific demethylase 1 (LSD1) inhibitors.
• To identify novel potent inhibitors of LSD1 using computational methods.

Main Methods:

• Constructed machine learning models using Morgan molecular fingerprints.
• Utilized a dataset of 931 molecules with known LSD1 inhibition activity from the ChEMBL database.
• Employed a support vector regressor, achieving an R2 of 0.703, for virtual screening.

Main Results:

• Identified five predicted potent LSD1 inhibitors from over 300,000 molecules in the ZINC database.
• Successfully recovered a known LSD1 inhibitor (RN1).
• Discovered four novel compounds with potential LSD1 inhibitory activity.

Conclusions:

• Machine learning-enabled virtual screening is effective for identifying LSD1 inhibitors.
• This study provides novel chemical entities for further development as anti-cancer agents targeting LSD1.