Supplementary Figure 2: Computational efficiency of SpatialDE.

Compared is the SpatialDE native implementation versus a Stan implementation of the same model. Caching operations and linear algebra speedups are used where possible, enabling tractable genome-wide analyses with thousands of samples or cells. Shown is the empirical runtime for the SpatialDE test applied to 10,000 genes, using a late 2013 iMac with 3.2 GHz Intel Core i5 processor.