Drugmonizome-ML

Drugmonizome-ML predicts novel drug indications and attributes by applying machine learning to annotated drug and small-molecule sets from the Drugmonizome database.

Key Features:

• Customizable Machine Learning Pipelines: Constructs tailored machine learning models using annotated drug and small-molecule sets from Drugmonizome.
• Integration with SEP-L1000 Phenotypic Datasets: Incorporates SEP-L1000 cellular phenotypic data to augment predictive features.
• Drug Set Libraries: Aggregates drug and small-molecule information including chemical structure, targets, indications, side effects, and induced gene expression signatures from public resources into structured libraries.
• Drug Set Enrichment Analysis: Performs enrichment analysis by querying the Drugmonizome database to identify biological processes and pathways associated with drug sets.
• Systematic Imputation of Properties: Applies machine learning techniques to impute novel properties and attributes for approved and preclinical compounds.

Scientific Applications:

• Drug Repurposing: Identifies potential repurposing opportunities by analyzing molecular and structural similarities across heterogeneous drug sets.
• Novel Property Discovery: Discovers previously unknown properties of preclinical small molecules via predictive modeling.
• Side Effect Prediction: Predicts potential adverse effects of approved and preclinical drugs, exemplified by peripheral neuropathy prediction.
• Consensus Enrichment Analysis of SARS-CoV-2 Screens: Performs consensus enrichment across 12 independent SARS-CoV-2 in vitro screens to identify common biological processes that inhibit viral replication.

Methodology:

Processes and abstracts information from public databases into structured drug set libraries, constructs machine learning models from Drugmonizome data, and performs drug set enrichment analysis and predictive modeling, including consensus enrichment analysis across multiple screens.