It came from another dimension!

Peter Todd will talk about visualising multidimensional data.

Biological data is messy — especially when it’s spatial. Large multi-channel images, point clouds, segmentation geometry, and tables (with associated sparse expression matrices) all need to work together in ways that support robustness and reproducibility.
In this talk, I’ll introduce the emerging SpatialData standard, built around a Zarr-based container. This organises chunked N-dimensional arrays alongside tabular and geometric data (Parquet, Arrow, and AnnData), linked through explicit coordinate system transformations and associated metadata. I’ll contrast this with more familiar tabular approaches (such as Parquet/Iceberg), and explore what changes when queries are spatial rather than purely analytical.
I’ll also demonstrate interactive visualisation in MDV, highlighting user-interface abstractions aimed at making exploration and presentation accessible across a range of technical backgrounds. More broadly, this is a discussion about how we build open, usable data ecosystems for complex scientific data — balancing specialisation and generalisation, fidelity and frugality, and correctness without sacrificing flexibility.