/Mhubai-IDC

Primary LanguageJupyter NotebookMIT LicenseMIT

MHUBAI Unleashed

Welcome to the mhubai-unleashed repository! This repository hosts WDL (Workflow Description Language) workflows for MHUBAI (Medical Hub AI), providing a structured and scalable way to run complex pipelines.

Please cite MHUBAI as this repo is based on their work: https://github.com/MHubAI

Overview

The workflows in this repository are designed to facilitate the execution of MHUBAI tasks, ensuring reproducibility and ease of use.

Getting Started

To get started with these workflows, follow these steps:

  1. Visit our Dockstore page: MHUBAI Unleashed on Dockstore

  2. Import the workflow to a platform of your choice:

  3. Provide the following inputs:

    • MHUBAI model name
    • aws or gcs urls in a txt or csv file, each line containing a single URL (choose one provider but not mix aws and gcp)
      • TIP: If files are downloaded from ImagingDataCommons, all SOPInstances in a series can be referenced by wildcard such as s3://idc-open-data/d0686cc8-0f8b-4e77-8707-605d1d8f7a08/*
    • If you want to override mhubai's default config, provide your custom config in yaml format while submitting the workflow
  4. A sample data table manifest you can import can be found in the repo