Drugmonizome

Drugmonizome aggregates and annotates publicly available small-molecule and drug data to build drug set libraries for drug set enrichment analysis and machine-learning prediction of drug properties and side effects.

Key Features:

• Data integration: Integrates publicly available data from web-based tools, databases, and repositories to collect small-molecule attributes.
• Attribute abstraction: Systematically processes, abstracts, and aggregates information on molecular structures, targets, indications, side effects, and induced gene expression signatures.
• Drug set libraries: Constructs comprehensive drug set libraries encapsulating annotated sets of drugs and small molecules.
• Drug set enrichment analysis: Performs drug set enrichment analysis to identify shared molecular and structural features and prioritize repurposing hypotheses.
• Consensus enrichment analysis: Enables consensus enrichment analysis across multiple in vitro screens, demonstrated on 12 independent SARS-CoV-2 screens.
• Machine learning pipelines: Provides Drugmonizome-ML pipelines that use the drug set libraries to train predictive models of drug properties and side effects.
• Side-effect prediction: Includes predictive modeling of side effects, exemplified by predicting whether approved and preclinical drugs induce peripheral neuropathy.
• Systems pharmacology resource: Integrates diverse drug-related data to support systems pharmacology investigations.

Scientific Applications:

• Drug repurposing: Identify novel repurposing opportunities by finding molecular and structural similarities among diverse drug sets.
• Side-effect prediction: Predict potential side effects of approved and preclinical compounds, such as peripheral neuropathy.
• Screen interpretation: Derive consensus biological processes from multiple in vitro screening experiments, as demonstrated with SARS-CoV-2 screens.
• Predictive model development: Develop machine-learning-based predictive models for drug effects using constructed drug set libraries.
• Systems pharmacology analyses: Support systems-level analyses to explore compound mechanisms, targets, and interactions.

Methodology:

Integrates public datasets, systematically processes and abstracts small-molecule attributes, aggregates them into drug set libraries, performs drug set enrichment and consensus enrichment analyses, and applies Drugmonizome-ML machine learning pipelines using those libraries.