Bioinformatics Seminar
Time: 11AM
Venue: Davis Auditorium and Online
2 June 2026
Spatial Analysis of Human Foetal Kidney and Kidney Organoids
Melody JinWEHI
Spatial transcriptomics enables gene expression profiling in situ, offering new opportunities to study cellular organization in complex tissues. My PhD develops spatially aware analysis methods and applies them to a complex kidney organoid dataset. The first aim was spatially aware marker gene analysis. We developed jazzPanda, a hybrid framework that identifies spatially relevant marker genes by using transcript and cell coordinates. The second aim extended this framework to gene set testing in spatial data, supporting two applications. The first application annotates clusters with cell-type-specific gene sets, and the second application tests for enrichment of biological pathways within cell types. The third aim applies these approaches to a large Xenium kidney dataset and will be the focus of the seminar. Using the Xenium platform with the 380-gene immuno-oncology panel, we profiled 28 human kidney organoids and 2 human foetal kidney samples. We investigate how loss of NRF2, a key oxidative stress regulator, affects development and injury responses in organoids. Our dataset spans over 1 million cells across multiple experimental conditions, including wildtype and knockout genotypes, two developmental stages, and varying oxygen conditions. We investigated several approaches for accurately annotating the cell types of the kidney organoids, including using the human foetal kidney samples as a reference dataset. We first manually annotated the human foetal kidney samples, testing spatially resolved clustering as well as single cell techniques. We applied jazzPanda to identify marker genes and used an independent single cell dataset to create gene sets to help annotate the clusters using our gene set testing approach. We then jointly clustered the foetal and organoid samples, which showed several shared cell types. However, a large number of cells were organoid specific and lacked their cell type counterparts in the human foetal kidney, prompting us to cluster the organoid samples independently and perform lineage-specific reclustering. This produced robust and reliable cell type identification across all organoid samples and enabled downstream analysis to compare cellular composition shifts and gene expression changes across genotypes, developmental stages and oxygen conditions.