Personal Cancer Genome Reporter (PCGR)

Patient-Specific Modeling in Python

cBioPortal is a data visualization tool for cancer genomics. This repository contains all components to deploy the customized version of cBioPortal used in the PM4Onco use case of the Medical Informatics Initiative Germany.

preon (PREcision Oncology Normalization) is a fuzzy search tool for medical entities.

A text-based computational framework for patient -specific modeling for classification of cancers. iScience (2022).

This repository contains the code and data related to the CIViC-Fact project

MIRACUM-Pipe is a whole exome and panel sequencing pipeline for precision oncology (https://github.com/AG-Boerries/MIRACUM-Pipe). This repository is a port of the pipeline running inside a singularity container.

Cancer prognosis with shallow tumor RNA sequencing

Reference-free detection of chromosomal breakpoints from NGS data. Identifies structural variations linked to cancer without reference genome dependency. Published in Bioinformatics (2014). Direct sample-to-sample comparison for precision oncology applications.

Research framework for POLE c.138del (p.Leu46Phefs*8); Confirmed germline pathogenic frameshift variant clinically consistent w/ PPAP (reported ultra-hypermutated; tumour signature confirmation pending). Mechanistic models, differential diagnosis, therapeutic strategies, and experimental priorities.

Computational phenotyping used in the paper "A Computational Pathology Model to Predict Docetaxel Benefit in Localized High-Risk and Metastatic Prostate Cancer"

💊 Advanced Drug Response Prediction & Multi-Omics Platform Interactive computational biology dashboard with ML integration, synthetic CCLE/GDSC data, dose-response modeling, and biomarker discovery. Features 6-tab Streamlit interface, Random Forest predictions, and publication-quality visualizations.

Designed and validated a computational oncology pipeline for hepatocellular carcinoma using RNA expression data, integrating machine learning, survival analysis, SHAP explainability, tumor microenvironment deconvolution, and multi-cohort external validation for precision medicine applications.

The repository can re-generate the learning curves, wall-clock times and policy heat-maps reported in the article: "Physics-guided deep reinforcement learning for personalized PDE control in cancer therapy optimization."

End-to-end multi-omics cancer subtype classifier XGBoost, pathway-aware deep fusion network (mRNA/miRNA/methylation/CNV), KEGG attention, SHAP/IG explainability, and a Streamlit demo. TCGA BRCA & COAD. PyTorch.

spora [bench] is a benchmark for spatial proteomics foundation models.

Fusion Oncology fuses XGBoost drug-sensitivity models with DNABERT-2 genomic embeddings, then routes predictions through digital twin simulation, PK/PD pharmacokinetics, GNN scoring, and Bayesian uncertainty to produce confidence-scored companion diagnostic reports.

Official implementation: A Clinical Informatics Framework for Myeloid Oncology. Integrates Scalable AI (GBDT, LSTM) and LLMs (Agentic AI) for precision oncology trial management, patient stratification, and automated FDA 21 CFR Part 11 compliance.

A statistical framework for detecting significant drug combination synergies in cancer. By leveraging tissue-specific reference null distributions across multiple synergy metrics, we compute empirical p-values to standardize synergy detection, uncover novel interactions, and enable rigorous evaluation of drug combinations.

Biological Hybrid AI pipeline for molecular subtyping of gastric cancer using multi-omics data (WES + DNA methylation + clinical). 91.2% accuracy, 100% MSI recall, 57.1% POLE recall.