Consent

Consent performs consensus ligand-based virtual screening to identify potential compounds that interact with specific protein targets by leveraging multiple known active molecules.

Key Features:

• Consensus queries: Implements consensus queries via serial screening with different known actives and by combining molecular descriptors into a consensus fingerprint.
• Chemical fingerprints: Supports MACCS (166 bits), ECFP4 (2048 bits), and an unfolded version of MOLPRINT2D.
• Datasets: Benchmarks methods on two datasets comprising 19 protein targets, 3,776 known active molecules, and approximately 2,000,000 inactive molecules.
• Consensus policies: Evaluates four different consensus policies across five consensus sizes.
• Scoring strategy: Ranks candidate molecules by the maximum score obtained against all known active molecules.
• Performance trade-offs: Demonstrates that consensus fingerprints can match screening performance when few actives are available and can provide superior computational speed in constrained scenarios.

Scientific Applications:

• Ligand-based virtual screening: Identification of potential active compounds for specific protein targets using multiple known actives.
• Consensus method benchmarking: Comparative evaluation of consensus policies and fingerprint types for virtual screening performance.
• Screening optimization: Selection of strategies that balance screening accuracy and computational efficiency.

Methodology:

Evaluates consensus methods on two experimental datasets using MACCS (166 bits), ECFP4 (2048 bits) and unfolded MOLPRINT2D fingerprints, benchmarks four consensus policies across five consensus sizes, and assesses ranking by the maximum score against known actives.