Spatial ploidy inference using quantitative imaging.

Abstract

Polyploidy (whole-genome duplication) is a common yet under-surveyed property of tissues across multicellular organisms. Polyploidy plays a critical role during tissue development, following acute stress, and during disease progression. Common methods to reveal polyploidy involve either destroying tissue architecture by cell isolation or tedious identification of individual nuclei in intact tissue. Therefore, there is a critical need for rapid and high-throughput ploidy quantification using images of nuclei in intact tissues. Here, we present iSPy (inferring Spatial Ploidy), an unsupervised learning pipeline that is designed to create a spatial map of nuclear ploidy across a tissue of interest. We demonstrate the use of iSPy in Arabidopsis, Drosophila, and human tissue. iSPy can be adapted for a variety of tissue preparations, including whole mount and sectioned. This high-throughput pipeline will facilitate rapid and sensitive identification of nuclear ploidy in diverse biological contexts and organisms.

Code and data availability

The paper deposits its paper-specific phenotyping assets publicly: confocal images of A. thaliana, D. melanogaster, and human cardiomyocytes, ilastik segmentation files, and A. thaliana cotyledon flow cytometry data are all in an OSF repository, and the iSPy analysis code is available both on OSF and in a public GitLab