Share with:

Image data resource – Practical

Trainer: Jean Marie Burel, Petr Walczysko and Frances Wong

Overview: 

This section illustrates how to access the IDR data using the Python API. As IDR is based on OMERO, the study about OMERO API is of advantage for this course. The main workflow in this practical fetches image data and metadata from IDR studies associated with a publication and re-analyses the data using an alternative segmentation tool to the one used by the publication authors. Also, the Python tooling is explored to achieve this. Further, the image (re)-analysis using deep learning analysis packages such as CellPose and StarDist are explored. The section uses Jupyter notebooks, an easily accessible web-based environment which makes working with code easier.

Learning outcomes:

After this section you should be able to

  • Explore IDR system of studies and linkage to publications programmatically
  • Identify how to find IDR images based on metadata and how to find associated ROIs/segmentations using OMERO API
  • Describe the metadata ecosystem inside IDR on the example of genes and phenotypes
  • Use the rich metadata in connection with images for reanalysis to verify the publication conclusions 
  • Know principles of how to programmatically link the IDR with other resources, such as Humanmine

Materials:

Other resources:

Additional information:

Should you wish to run the practical in your local computer you can do so by opting for one of the following two options:

Option 1: Use our notebooks provided by the trainers in Google Colab during the course, which does not require any pre-installed software, only any active Google account.

Option 2: Install the Mamba cross-platform package manager on your computer, which can be completed by:

1. If you do not have any pre-existing conda installation, please install Mamba the recommended way, this means using mambaforge. 

2. In case you have a pre-existing conda installation, you can install Mamba by either:

 2a) Using the recommended way to install Mamba from https://github.com/conda-forge/miniforge#mambaforge  This will not invalidate your conda installation, but possibly your pre-existing conda environments will be in a different location (e.g. “/Users/user/opt/anaconda3/envs/“) then the new mamba environments (e.g. “/Users/user/mambaforge/envs/“). You can verify this by running “conda env list“. The addition of a line “export CONDA_ENVS_PATH=/Users/user/opt/anaconda3/envs/“ into your “.bashprofile“ or “.zprofile“ file will fix this.

  2b) Using the Existing conda install way, i.e. run “conda install mamba -n base -c conda-forge“ whilst in the base environment. This way can take much longer time than the recommended way described above, and might not lead to a successful installation, especially if run on arm64 (Apple Silicon) OS X.

For communication: https://forum.image.sc/