Related Experiment Video
Updated: Jun 22, 2025
A Combined 3D Tissue Engineered In Vitro/In Silico Lung Tumor Model for Predicting Drug Effectiveness in Specific Mutational Backgrounds
Published on: April 6, 2016
CODEX: COunterfactual Deep learning for the in silico EXploration of cancer cell line perturbations
Stefan Schrod1, Helena U Zacharias2, Tim Beißbarth1
1Department of Medical Bioinformatics, University Medical Center Göttingen, 37077 Niedersachsen, Germany.
Motivation:
High-throughput screens (HTS) provide a powerful tool to decipher the causal effects of chemical and genetic perturbations on cancer cell lines. Their ability to evaluate a wide spectrum of interventions, from single drugs to intricate drug combinations and CRISPR-interference, has established them as an invaluable resource for the development of novel therapeutic approaches. Nevertheless, the combinatorial complexity of potential interventions makes a comprehensive exploration intractable. Hence, prioritizing interventions for further experimental investigation becomes of utmost importance.
Results:
We propose CODEX (COunterfactual Deep learning for the in silico EXploration of cancer cell line perturbations) as a general framework for the causal modeling of HTS data, linking perturbations to their downstream consequences. CODEX relies on a stringent causal modeling strategy based on counterfactual reasoning. As such, CODEX predicts drug-specific cellular responses, comprising cell survival and molecular alterations, and facilitates the in silico exploration of drug combinations. This is achieved for both bulk and single-cell HTS. We further show that CODEX provides a rationale to explore complex genetic modifications from CRISPR-interference in silico in single cells.
Availability And Implementation:
Our implementation of CODEX is publicly available at https://github.com/sschrod/CODEX. All data used in this article are publicly available.
Insights
High-throughput screens (HTS) identify cancer drug effects, but complexity limits exploration. CODEX (COunterfactual Deep learning for in silico EXploration) uses causal AI to predict drug responses and combinations, enabling efficient in silico analysis.
Area of Science:
- Computational biology
- Cancer research
- Machine learning
Background:
- High-throughput screens (HTS) are crucial for understanding cancer cell line responses to chemical and genetic perturbations.
- The vast combinatorial possibilities of interventions (drugs, combinations, CRISPR) make comprehensive HTS exploration challenging.
- Prioritizing interventions is essential for efficient therapeutic development.
Purpose of the Study:
- To introduce CODEX (COunterfactual Deep learning for the in silico EXploration of cancer cell line perturbations), a framework for causal modeling of HTS data.
- To enable in silico prediction of drug-specific cellular responses, including survival and molecular alterations.
- To facilitate the exploration of drug combinations and genetic modifications using counterfactual reasoning.
Main Methods:
- Developed CODEX, a general framework for causal modeling of HTS data.
- Employed counterfactual reasoning for stringent causal modeling.
- Applied the framework to both bulk and single-cell HTS data.
- Utilized deep learning for in silico prediction of perturbations' effects.
Main Results:
- CODEX accurately predicts drug-specific cellular responses and molecular alterations.
- The framework enables in silico exploration of synergistic and antagonistic drug combinations.
- CODEX provides a rationale for exploring complex genetic modifications, such as CRISPR-interference, in single cells.
- The approach is effective for both bulk and single-cell HTS data.
Conclusions:
- CODEX offers a powerful computational tool for prioritizing and exploring interventions in cancer research.
- The framework leverages causal AI and counterfactual reasoning to overcome the limitations of HTS combinatorial complexity.
- CODEX facilitates the discovery of novel therapeutic strategies by enabling efficient in silico experimentation.
