Skip to main content
PantheonDAG Markey Cancer Center, University of Kentucky

A Markey Cancer Center project

Explore your data.

Ask a question. Build an analysis.

PantheonDAG is an AI assistant for research analysis. Upload a dataset to your workspace, give its full path in the chat, and describe your question in plain language. The assistant writes and runs analysis code in your private workspace on campus. Results, figures and notebooks are saved in your account for review.

Research workflows

What PantheonDAG can do

The pilot includes a single-cell analysis workflow that records how analysis steps connect. The underlying software's documentation also covers these kinds of work:

  • Single-cell and spatial genomics

    Analyze single-cell and spatial genomics datasets using the documented workflows.

  • Sequencing data preparation

    Process FASTQ sequencing files with the documented processing pipeline.

  • Data analysis and visualization

    Summarize tables, run statistical tests and regression, explore machine learning and time series, and create plots using Python or R.

  • Research literature

    Analyze and summarize academic papers, then check the summaries against the original sources.

Available workflows depend on the workspace's installed software and configuration. Discuss your analysis needs with the project contacts.

Read before you sign in

Using AI responsibly

Language models can produce inaccurate, biased or invented answers, including incorrect code and citations. Check results against reliable sources, test generated code, and review every action before relying on it. You remain responsible for your research, its conclusions, and appropriate disclosure of AI assistance.

Use this service for research and academic work only. It is not a substitute for professional judgment or a service for commercial or unrelated personal use.

Keep sensitive information out

Do not submit protected health information (PHI/HIPAA), personally identifiable information, FERPA student records, export-controlled information, Controlled Unclassified Information (CUI), passwords, or API keys in prompts, datasets or shared workspaces. Campus hosting does not authorize sensitive or regulated data.

01 / Access

One institutional sign-in

Sign in with your University of Kentucky linkblue account through CILogon. Workspace access also requires a CCS-approved portal account.

02 / Model access

A default key is provided. Your own key is optional.

PantheonDAG provides approved accounts with a shared site key for the campus inference gateway operated by CCS. It is stored privately in your account. You can read it there, so treat it as a shared credential: do not copy it into notebooks, source repositories, messages or tickets. CCS manages its rotation. Your own Locksmith key stays under your control.

Use your own Locksmith key with only a browser
  1. Sign in at CCS Key Management (Locksmith). If needed, ask the CCS service desk to enable inference access. An SSH key is not required for model access.
  2. Open AI Inference Keys, generate your personal key, and copy it privately.
  3. In this portal, open Files → Home Directory. Enable Show Dotfiles (hidden files), open .pantheondag, then use Edit on .env. If it is missing, copy env.example to .env in the same directory.
  4. Replace both LLM_API_KEY and OPENAI_API_KEY with your personal key. Keep both base URLs at https://llm.ccs.uky.edu/v1. Use one NAME=value per line and save the file.
  5. Delete any running session and launch a new one to load the change. Open Help → Getting started for the full walkthrough.

Files editing is available after portal approval. If you cannot open or save the private file, contact CCS; never send the key in a support request.

03 / Trust & privacy

Responsible access to PantheonDAG

This University of Kentucky system is for authorized use. Access and operational activity may be recorded for security and service administration. Unauthorized use is prohibited and may carry disciplinary, civil or criminal consequences.

Check the selected model's hosting and data handling before use. Do not assume prompts, files or results are confidential merely because you signed in. Follow university policies and IT security guidance.

04 / Support

Hosting, sign-in and model access support

PantheonDAG is a Markey Cancer Center project at the University of Kentucky. The Center for Computational Sciences provides its hosting, CILogon sign-in and on-campus model inference. Contact CCS for help with this infrastructure.

Open a CCS service desk request
Read the Locksmith key guide
Read gateway help and error explanations

Include the time, action and error message. Remove credentials, sensitive data and personal identifiers from screenshots.

About the project

Project contacts