Multiplexed 3D atlas of state transitions and immune interactions in colorectal cancer (Minerva Abstract)

The collection comprises preservation-quality files of Minerva output files without needing download terabyte scale images. To view data in a browser follow this link: https://www.cycif.org/data/lin-wang-coy-2021/viz.html. This dataset uses the Minerva Suite, a series of software tools developed by the Laboratory of Systems Pharmacology to visualize multiplexed tissue image data in a web browser. Researchers or pathologists can annotate and describe images for users and users can use zooming and panning features to explore the full resolution images without needing to download multi-GB/TB image files. These annotated and unannotated images are created by uploading quality controlled ome.tiffs and segmentation masks, along with channel metadata and text descriptions, into the Minerva Author tool. These input files produce .json files, an .html file, and hundreds of .jpg pyramid files that make the images browsable online. This dataset uses multiplexed tissue imaging, spatial statistics, and machine learning to identify cell states underlying morphological features of known prognostic significance in colorectal cancer. We find that the necessary spatial analysis requires extended tumor domains, not tissue microarrays or small view-fields. When this condition is met, the data reveal frequent transitions between histological archetypes (tumor grades and morphologies) correlated with molecular gradients. At tumor invasive margins, where tumor, normal, and immune cells compete, localized features in 2D such as tumor buds and mucin pools are seen in 3D to be large connected structures having continuously varying molecular properties. Immunosuppressive cell-cell interactions also exhibit graded variation in type and frequency. Thus, whereas scRNA-Seq emphasizes discrete changes in tumor state, whole-specimen imaging reveals the presence of multi-scale spatial gradients analogous to those in developing tissues.