Predicting synergistic effects between compounds through their structural similarity and effects on transcriptomes

Area of Science:

• Computational biology
• Pharmacology
• Bioinformatics

Background:

• Combinatorial therapies are vital for cancer treatment but face challenges due to the vast number of potential drug combinations.
• Exhaustive screening of all possible combinations is computationally prohibitive, necessitating predictive tools.

Purpose of the Study:

• To develop computational methods for predicting compound combination effects and prioritizing synergistic drug pairs.
• To identify features that are informative about drug synergy using the NCI-DREAM Drug Synergy Prediction Challenge dataset.

Main Methods:

• Systematic exploration of differential gene expression profiles after single compound treatments.
• Comparison of molecular structures of candidate compounds.
• Statistical analysis to associate feature types with experimentally measured combination effects.

Main Results:

• Drug combinations exhibiting synergistic effects were significantly associated with compounds having dissimilar molecular structures.
• Similarity in induced gene expression changes between compounds also correlated with synergy.
• These two feature types (structural dissimilarity and expression similarity) provide complementary information for synergy prediction.

Conclusions:

• Insights into the mechanisms underlying drug combination effects were gained.
• The identified features can help prioritize promising drug combinations within a large search space, accelerating combinatorial therapy development.