How well do Pyxa 3D cell types agree with scRNA-seq reference data (235-plex Mouse Neuron Cell Typing panel)?

We validated the 235-plex Mouse Neuron Cell Typing panel against the Allen Brain Cell Atlas (whole mouse brain scRNA-seq), which is a widely used reference for mouse brain cell types.

For each segmented cell in a mouse brain section, we calculated the Spearman correlation between its expression profile and the Allen Brain Atlas subclass reference profiles. Most cells show correlations of roughly 0.7 to 0.8 with their matched subclass, and the correlation map follows the expected anatomy. Regions with lower correlation (0.4 to 0.6) are limited and localized.

This means Pyxa’s spatial measurements agree well with scRNA-seq-defined cell identities, using only 235 genes and without a whole-transcriptome readout. Please note that the correlation is calculated on the panel genes only. It is a measure of concordance with the reference, not a per-gene accuracy metric. Lower correlation in a region can reflect true biology, such as cell types or states not covered by the panel or reference, or technical factors like segmentation or tissue quality.