MetaComb: Drug Combination Response Predictions for Cell Lines and Patients

A meta-learning framework for predicting drug combination responses, developed by Congcong Guo and last updated in June 2026. It uses drug structures and gene expression profiles from cell lines and patient ex vivo samples as input. The method was trained on data-rich cell line tasks and fine-tuned for scenarios with limited patient-derived samples.

Use Cases

• Predicting patient ex vivo drug combination response based on drug structures and gene expression profiles.
• Fine-tuning meta-learning models for few-shot prediction tasks in oncology.
• Benchmarking meta-learning approaches against conventional transfer learning methods for drug response prediction.

Strengths

• The MetaComb framework improved AUROC by 8.5% for data-poor cell lines and by 7.4% for patient ex vivo samples compared to conventional transfer learning.
• The dataset is shared under a CC-BY-4.0 license, permitting open reuse with attribution.
• The description provides a detailed methodological and results framework for the prediction model.

Limitations

• The dataset is very small at 15.0 KB, indicating limited scope.
• Row count and column-level documentation are unknown, which limits suitability assessment.
• The authors note that current patient-level assessment is insufficient to support generalization to other cancer types or patient populations.

Provenance

Source

Congcong Guo via figshare.

Collection Method

Likely contains results from a meta-learning framework (MetaComb) trained on cell line data and fine-tuned on patient ex vivo samples.

Freshness

Last updated 2026-06-01 22:02:56; freshness should be verified.