Relating drug-protein interaction network with drug side effects.

Area of Science:

• Pharmacology and Bioinformatics
• Computational Drug Discovery
• Systems Biology

Background:

• Identifying drug side effects and their mechanisms is crucial but challenging in drug development.
• System-wide approaches are needed to link molecular-scale drug-protein interactions with phenotypic-scale side effects for prediction.
• This study addresses the need for better prediction of side effects for uncharacterized drugs.

Purpose of the Study:

• To develop and validate a method for predicting drug side effects by linking drug-protein interactions with phenotypic side effect profiles.
• To identify biologically relevant relationships between drug targets and their associated side effects.
• To facilitate the prediction of potential side effects for new drug candidates.

Main Methods:

• Performed a large-scale analysis using sparse canonical correlation analysis to extract correlated sets of targeted proteins and side effects.
• Analyzed data from 658 drugs, 1368 proteins, and 1339 side effects to identify 80 correlated sets.
• Utilized KEGG and Gene Ontology enrichment analyses to interpret the biological relevance of the correlated sets.

Main Results:

• Successfully extracted 80 correlated sets linking drug-targeted proteins to specific side effects.
• Enrichment analyses revealed that correlated sets were significantly enriched with proteins in the same biological pathways, supporting biological relevance.
• Demonstrated that extracted side effects represent potential phenotypic outcomes of targeting proteins within the same correlated set.

Conclusions:

• The proposed method effectively links molecular-scale drug-protein interactions with phenotypic-scale side effects.
• The findings provide a biologically relevant interpretation of drug-target-side effect relationships.
• The method is valuable for predicting potential side effects of new drug candidates based on their protein-binding profiles.