lamm-mit/skills/bindingdb-database

BindingDB Database Skill Summary

Overview

BindingDB is a major public repository containing "over 3 million binding data records for ~1.4 million compounds tested against ~9,200 protein targets." The database stores quantitative binding measurements essential for pharmaceutical research and computational chemistry.

Primary Use Cases

This resource excels when researchers need to:

• Identify known compounds that bind to specific protein targets
• Conduct structure-activity relationship (SAR) analyses examining how molecular modifications impact binding strength
• Assess compound selectivity across multiple protein targets
• Source curated affinity datasets for machine learning applications
• Evaluate potential off-target binding for drug repurposing studies

Core Query Methods

The skill provides multiple access approaches: REST API queries by UniProt target ID, compound name searches, SMILES-based lookups, and large-scale TSV file downloads for comprehensive analysis.

Key Measurement Types

BindingDB tracks four primary affinity metrics: Ki (inhibition constant), Kd (dissociation constant), IC50 (half-maximal inhibition), and EC50 (half-maximal effectiveness). Values below 10 nM typically indicate drug-potency compounds.

Data Quality Considerations

Results should be filtered by target organism to ensure human protein relevance, and users should recognize that IC50 values vary based on experimental conditions like substrate concentration.