WADDAICA: A webserver for aiding protein drug design by artificial intelligence and classical algorithm

Area of Science:

• Computational chemistry
• Drug discovery
• Bioinformatics

Background:

• Artificial intelligence (AI) and classical algorithms offer distinct advantages in drug design.
• Deep learning models excel with large datasets, while classical algorithms follow predefined procedures.
• Integrating both approaches can enhance drug design efficiency and novelty.

Purpose of the Study:

• To develop a webserver, WADDAICA, combining deep learning and classical algorithms for drug design.
• To provide tools for scaffold hopping and drug modification using AI and established methods.
• To offer a user-friendly platform for researchers to design novel drug candidates.

Main Methods:

• WADDAICA employs deep learning models for scaffold hopping, trained on PDBbind database.
• Classical algorithms are integrated for drug modification and design.
• The webserver provides a convenient interface for job submission and analysis.

Main Results:

• The deep learning model demonstrates strong scoring power for compound evaluation.
• WADDAICA's classical algorithms show superior Pearson and Spearman correlations in binding affinity prediction compared to Autodock Vina.
• The integrated platform facilitates efficient drug design and modification.

Conclusions:

• WADDAICA effectively leverages both deep learning and classical algorithms for drug design.
• The webserver offers a valuable and accessible tool for researchers seeking to discover novel drugs.
• WADDAICA enhances drug discovery pipelines through its combined computational approaches.