Fig. 2

Decoding spatial maps of layer-specific excitatory neurons in the human brain. A The composition of ten excitatory neuron types in the human brain was predicted by spSeudoMap. The estimated cell fraction was spatially mapped to the tissue and visualized with colored bars. Overall, the spatial distribution of layer-specific neuron types was restricted to the corresponding cortical layer (layers 1 to 6). B The performance was measured by calculating the layer discriminative accuracy of the predicted layer-specific neuron fraction. It was represented by the area under the receiver operating characteristic curve (AUROC). The AUROC values of spSeudoMap in 10 cell subtypes were visualized with barplots and compared with those of CARD, CellDART, Cell2location, DSTG, RCTD, and SPOTlight. All methods were implemented using the simulated single-cell data of excitatory neuron subpopulations. In contrast to other methods, spSeudoMap showed a more stable performance represented by AUROC over 0.5 across all neuron types