Transcriptomic correlates of electrophysiological and morphological diversity within and across excitatory and inhibitory neuron classes
Fig 5
Assessing gene-property relationships within cell subclasses using PatchSeq (A) Number of genes associated with each electrophysiological property based on a joint cross-laboratory analysis of 5 PatchSeq datasets. Genes shown are significant at FDR = 0.1, based on a mixed-effects regression model, treating gene expression as a fixed effect and dataset identity and cell type as random effects. Bar color denotes overlap of PatchSeq based gene-property relationships with analogous relationships from the AIBS class-conditional model analysis. Note that analysis of gene-property relationships in the PatchSeq datasets are independent from those in the AIBS cell types analysis. (B, C) Examples of genes showing significant associations with electrophysiological features in the class-conditional analysis of the AIBS dataset (left-most panel) and the mixed-effects analysis of the PatchSeq datasets (other panels). Dataset name and cell type is shown in the subpanel title and solid lines indicate linear fits within cell classes (AIBS) or fits within each PatchSeq dataset and cell type, after weighting cells by transcriptome-quality (see Methods). Based on differences in mRNA quantification, x-axis units for AIBS, Cadwell, and Földy datasets are log2 (CPM+1), and for Bengtsson Gonzales, Muñoz, and Fuzik datasets are log2 normalized molecule counts (normalized to 2000 unique molecules per cell).