Integration of Computational Docking into Anti-Cancer Drug Response Prediction Models.

Area of Science:

• Oncology
• Computational Biology
• Pharmacology

Background:

• Cancer exhibits heterogeneity, leading to varied responses to treatments even among tumors of the same histology.
• Accurate anti-cancer drug response prediction is crucial for effective drug development and personalized patient treatment strategies.
• Current computational models face challenges due to the complex mechanisms of cancer and drug interactions.

Purpose of the Study:

• To investigate the integration of computationally derived molecular mechanism of action features into anti-cancer drug response prediction models.
• To assess the feasibility of enhancing prediction accuracy by incorporating molecular docking scores alongside gene expression and drug descriptors.

Main Methods:

• Utilized cancer gene expression data and molecular drug descriptors.
• Integrated computationally derived docking scores between drug molecules and target proteins.
• Developed and tested response prediction models using these combined features on large-scale drug screening data.

Main Results:

• A marginal improvement in anti-cancer drug response prediction performance was observed when docking scores were included as additional features.
• The study established a baseline dataset of large-scale computational docking for anti-cancer drugs.

Conclusions:

• Integrating molecular docking information shows potential for enhancing computational drug response prediction models.
• Further research is needed to overcome limitations and refine the approach for more significant predictive power.