MdrDB

MdrDB integrates experimental and structural data to characterize how mutations alter protein-ligand binding affinities (ΔΔG) and support studies of mutation-induced drug resistance.

Key Features:

• Integration of Multiple Datasets: Integrates data from seven publicly available datasets to assemble a unified resource focused on mutation-induced drug resistance.
• Extensive Data Coverage: Contains 100,537 samples covering 240 proteins, 5,119 Protein Data Bank (PDB) structures, 2,503 mutations, and 440 drugs.
• Detailed Sample Information: Provides 3D structures of wild-type and mutant protein-ligand complexes, binding affinity changes upon mutation (ΔΔG), and biochemical features for each sample.
• Enhanced Predictive Capabilities: Experimental results show MdrDB improves the performance of commonly used machine learning models for predicting ΔΔG across three standard benchmarking scenarios.

Scientific Applications:

• Mechanistic analysis of drug resistance: Enables analysis of molecular mechanisms by which mutations affect drug binding through structural comparisons and ΔΔG data.
• Integration with drug sensitivity datasets: Supports integration of drug sensitivity and cell line mutation data from Genomics of Drug Sensitivity in Cancer and DepMap for comprehensive analyses.
• Machine learning model development and benchmarking: Facilitates training and benchmarking of models to predict mutation-induced changes in binding affinity (ΔΔG).

Methodology:

MdrDB was constructed by integrating seven publicly available datasets and compiling protein structures, mutations, drug interactions, ΔΔG values, and biochemical features into a unified dataset.