Gene prioritization seeks to identify the genes most relevant to the development or prevalence of a disease. This project consolidates our group's existing gene-prioritization methods into a single, user-friendly system. The focus is on software architecture, a clean package, a simple web interface, and a fair benchmarking workflow. Users can upload data, run comparable experiments with shared settings, and receive ranked results with clear summaries and explanations.
Objectives:
- Provide a unified interface and consistent inputs/outputs across methods.
- Offer a transparent benchmarking pipeline with shared splits, standard metrics, and clear reports.
- Delivers explainable results at both overview and per-gene levels.
- Ensure reproducibility through configuration files, version control, and containerized runs.
- Address ethical and responsible AI considerations (data provenance, leakage checks, non-clinical use).
System Outline:
A modular package handles data validation, feature preparation, method execution, evaluation, and report generation. A lightweight web layer triggers runs, tracks progress, and presents leaderboards, rankings, and explanations. All components are documented, tested, and designed for future extension.
Deliverables:
- Installable software package with command-line tools and documentation.
- Web demo with an example dataset and saved runs.
- Benchmarking report comparing current methods under identical conditions.
- Thesis describing design choices, evaluation results, and ethical reflections.
The project covers the full IT system life cycle—from analysis and design to implementation, use, and further development—while emphasizing openness, precision, and ethical awareness. It demonstrates professional competence in programming and systems architecture, clear communication of scientific results, and the ability to apply knowledge in a broader social and academic context.
If you want to learn more or get involved, please contact Pooya Zakeri or Ziyu Zhang, .