A Systems-Level Machine Learning Approach Uncovers Therapeutic Targets in Clear Cell Renal Cell Carcinoma

---

Overview

Ruhrberg Estévez, S., Baltusyte, G., Youssef, G. et al. npj Drug Discov. 3, 5 (2026).

We present a generalisable, interpretable machine learning framework for therapeutic target discovery using single-cell transcriptomics, protein interaction networks, and drug proximity analysis. The pipeline integrates feature selection via gradient boosting classifiers, systems-level network inference, and in silico drug repurposing, enabling the identification of actionable targets with cellular specificity. As a proof of concept, we apply the method to clear cell renal cell carcinoma (ccRCC), an aggressive kidney cancer with limited treatment options. The model identifies 96 tumour-intrinsic genes, refines them to 16 targets through CRISPR screens and biological curation, and prioritises FDA-approved compounds via network-based proximity scoring. Several novel therapeutic mechanisms — including ABL1, CDK4/6, and JAK inhibition — emerge from this analysis, with predicted compounds showing superior efficacy to standard-of-care drugs across multiple ccRCC cell lines. Beyond ccRCC, this framework offers a scalable strategy for drug discovery across diverse diseases, combining machine learning interpretability with systems biology to accelerate therapeutic development.

---

Repository Structure

├── 1_data_preprocessing/     # Data loading and annotation (US, Chinese, Harvard, metastatic datasets)
├── 2_feature_selection/       # UMAP visualisation, multiclass/single-class classifiers, univariate selection
├── 3_feature_validation/      # Validation across independent cohorts + ROC-AUC analysis
├── 4_drug_target_validation/  # Drug target validation across cohorts
├── 5_network_analysis/        # Protein interaction network construction, permutation tests, drug proximity
│   ├── databases/             # Network databases (PPI, drug-target)
│   └── src/                   # Network analysis source code
├── 6_drug_validation/         # Hallmark gene sets, z-values, preclinical trial analysis
├── data/                      # Gene lists, model files, predictions
├── output/                    # Generated figures (ROC curves, UMAPs)
├── rcc.yml                    # Conda environment specification
└── overview.png               # Pipeline overview figure

---

Getting Started

Prerequisites

• Conda or Mamba
• R (for data preprocessing scripts in 1_data_preprocessing/)

Environment Setup

Create and activate the conda environment:

conda env create -f rcc.yml
conda activate rcc

Note: The network analysis module (5_network_analysis/) optionally uses graph-tool, which must be installed separately via conda-forge:

conda install -c conda-forge graph-tool

Running the Pipeline

The pipeline is designed to be run in numbered order:

1. Data Preprocessing (1_data_preprocessing/) — Process raw scRNA-seq data and annotate datasets
2. Feature Selection (2_feature_selection/) — Train classifiers and perform univariate feature selection
3. Feature Validation (3_feature_validation/) — Validate selected genes across independent cohorts
4. Drug Target Validation (4_drug_target_validation/) — Validate drug targets across cohorts
5. Network Analysis (5_network_analysis/) — Build PPI networks, run permutation tests, compute drug proximity
6. Drug Validation (6_drug_validation/) — Hallmark enrichment, z-score analysis, preclinical validation

---

Key Dependencies

Package	Purpose
lightgbm	Gradient boosting classifier for feature selection
scikit-learn	Machine learning utilities and model evaluation
shap	Model interpretability via SHAP values
umap-learn	Dimensionality reduction for visualisation
networkx	Protein interaction network analysis
mygene	Gene annotation and ID mapping
pyomo	Optimisation for network proximity analysis
scipy / statsmodels	Statistical testing
seaborn / matplotlib	Plotting and visualisation

---

Citation

If you use this code, please cite:

@article{ruhrberg2026systems,
  title={A systems-level machine learning approach uncovers therapeutic targets in clear cell renal cell carcinoma},
  author={Ruhrberg Est{\'e}vez, S. and Baltusyte, G. and Youssef, G. and others},
  journal={npj Drug Discovery},
  volume={3},
  pages={5},
  year={2026},
  publisher={Nature Publishing Group},
  doi={10.1038/s44386-025-00036-z}
}