Link Search Menu Expand Document

Michigan Data Interface

The Michigan Data Interface (MDI) is a standardized framework for developing, installing and running HPC data analysis pipelines and interactive R Shiny visualization applications.

These pages describe the MDI project overall. Links within lead you to documentation for specific components of the MDI, e.g., how to get started writing your own tool suites.


Stage 1 pipelines via command line, YAML-based tool and job definition

Pipeline Runner web app running Stage 1 pipelines, remote HPC access

Stage 2 interactive apps via a access-controlled public web server

Quick Start

To see this live demo app server in action:

Follow the instructions at this link to clone and run the helper script that will install the MDI frameworks and your tool suites of interest.

Alternatively, use the following web site to generate a friendly, customized batch script for your desktop computer, e.g., to install and run the MDI remotely.

Tool developers should start by copying our repository suite template.

Guiding Principles

Principles guiding development of the MDI are:

  • easy, standardized implementation
  • simple, effective use of modern development tools
  • efficient use of scalable computation resources
  • maximum developer flexibility
  • rapid collaboration and code sharing
  • interactive data analysis

Our goal is to help you develop and share robust data analysis tools more easily, or to use tools developed by others, without forcing too many requirements into the process.

Basic Training for Developers

This documentation assumes familiarity with open source code development. Please see our Basic Training tutorials in the link below if you need help getting started with Git, R, Shiny, job schedulers, etc.

Portability and Licensing

As the name implies, the MDI was created to support researchers at the University of Michigan; a few instructions are specific to our environment. However, the MDI codebase is generic and can be used by any laboratory or organization for any data analysis need under the MIT license.

For more information about the MDI, please visit:

MDI Documentation MDI Basic Training MDI GitHub page