Predicting High-Impact Pharmacological Targets by Integrating Transcriptome and Text-Mining Features.

Area of Science:

• Oncology
• Pharmacology
• Bioinformatics

Background:

• Drug development efficiency has stagnated despite technological advancements.
• Novel strategies are crucial for evaluating candidate molecules in oncology.
• Current methods struggle to identify high-impact therapeutic targets effectively.

Purpose of the Study:

• To test the hypothesis that economic impact of targeting a gene product is predictable a priori.
• To develop a method combining transcriptome profiles and literature metrics for target impact assessment.
• To improve the efficiency of identifying promising anti-cancer drug targets.

Main Methods:

• Compared gene expression patterns of high-impact and low-impact anti-cancer targets.
• Used publication counts and their derivatives as metrics for public interest and bias minimization.
• Combined gene expression features with literature mining metrics for target classification.

Main Results:

• Achieved a 4-fold enrichment in high-impact targets using the combined approach.
• Demonstrated significant differences in economic value between high and low-impact targets.
• Identified known high-impact targets (e.g., EGFR, MTOR) and promising research-stage targets (e.g., p53, c-Jun).

Conclusions:

• A priori impact assessment of targets and ligands can significantly reduce drug development costs.
• Combining gene expression and literature data improves prediction of target economic impact.
• Expanding combinational treatments can increase impact per ligand and decrease overall costs.