Spatial transcriptomics defines injury-specific microenvironments in the adult mouse kidney and novel cellular interactions in regeneration and disease

Authors: Michal Polonsky, Louisa M. S. Gerhardt, Jina Yun, Kari Koppitch, Katsuya Lex Colón, Henry Amrhein, Barbara Wold, Shiwei Zheng, Guo-Cheng Yuan, Matt Thomson, Long Cai, Andrew P. McMahon

Introduction

The purpose of the Celltiles software development project at the Caltech Cell Interaction Initiative is to enable end-user biologists to interactively view and explore spatial transcriptome, proteome, and other spatial genomic datatypes, together with pertinent computed metadata, mapped to the source microscopic image coordinates.

Here, Celltiles version 1.0 is provided, together with example data and links to the full kidney injury data of Polonsky et al, 2024 (10.1038/s41467-024-51186-z) to provide viewing of the expression data, cell types, and other spatial analysis such as computed microenvironments (MEs).

Celltiles 1.0 allows a user to view large microscopy images with the familiar pan-and-zoom interface, inspired by online mapping software. Additionally, it allows overlaying categorical and quantitative annotation data (e.g. expression levels per gene) onto the images. The libraries used to create this software are open-source and can be freely distributed.

The images, as acquired from uManager, are combined into a stitched overview of the entire tissue, which is then disassembled into small tiles by software designed for processing geospatial imagery and stored in a hierarchical format on the backing file system. This facilitates requesting the images on an as-needed basis, rather than attempting to store the entire image in system memory.

Commonly used annotations of Seq-FISH and serial-FISH data include clustering of cells to identify their type and/or state (normal, diseased, or injured). The clustering itself is typically from either the Seurat package in R or from Scanpy in Python, although Celltiles is agnostic to the clustering package and parameters used to assign cluster membership and cell labels. These data are converted into JSON files to facilitate transferring the data into a Javascript application running in a web browser (Firefox, Chrome, Safari).

A containerized version of the Celltiles application is provided to allow one to use it locally if desired, including on a laptop without access to the Internet. One goal in making Celltiles available to the community at large is to stimulate discussion and advance further development.

Links

Acknowledgements

Celltiles Author: Henry Amrhein <hamrhein@caltech.edu>
California Institute of Technology
Barbara Wold Lab
Cell Interaction Initiative