The polypharmacology browser: a web-based multi-fingerprint target prediction tool using ChEMBL bioactivity data.

Area of Science:

• Computational chemistry
• Cheminformatics
• Drug discovery

Background:

• Existing web tools predict small molecule targets using limited molecular fingerprints (fps).
• Target prediction accuracy varies significantly based on the similarity computation method used.

Purpose of the Study:

• To develop a web-based platform for predicting small molecule targets using multiple molecular fingerprints simultaneously.
• To improve the accuracy and reliability of polypharmacology predictions.

Main Methods:

• Developed the polypharmacology browser (PPB) utilizing ten diverse molecular fingerprints (fps).
• PPB searches ChEMBL database for compounds with similar molecular fingerprints to predict targets.
• Employs a consensus voting scheme and p-value for ranking predicted targets.

Main Results:

• Validation across 670 drugs showed combining 10 fps improved prediction accuracy.
• Achieved 50% prediction of known drug targets with a 25% hit rate.
• Case study demonstrated PPB accurately predicted inhibition of a TRPV6 inhibitor against 5/5 targets.

Conclusions:

• The polypharmacology browser (PPB) provides a versatile platform for multi-fingerprint target prediction.
• PPB supports drug discovery by offering enhanced accuracy through simultaneous fingerprint comparisons.
• The tool is freely accessible for researchers.