2. Image-based profiling

Images contain a lot of information; if we use computer vision and deep learning to analyse images of cells, we can define quantitative profiles that contain a comparable amount of information to molecular profiling techniques like transcriptomics or metabolomics. The big benefit of image-based profiling is that it is much, much cheaper than these other techniques. Our group works a lot with the Cell Painting assay, which is a type of high-content imaging designed to jointly maximise information content and throughput.

Here are some key papers on the Cell Painting assay:

• Initial protocol introducing the Cell Painting assay (paper link)
• Updated protocol 7 years later (paper link)
• Analysis of image-based profiles (paper link)
• Pycytominer Python package for image-based profiling (paper link)
• Cell Painting Gallery (paper link)
• A review of Cell Painting applications (paper link)

To learn how to use CellProfiler to process Cell Painting images, run through the beginner segmentation CellProfiler tutorial following this recording. You can either do this locally or on the Codon Cluster by using the Xfce Desktop Environment available through the iHPC portal. In the portal, go to 'Interactive Apps', select 'Desktop', and launch a session. You should be able to install CellProfiler and download tutorial materials just like on your local computer.