Fig. 3

Independent validation of results. A Validation bulk RNA-seq datasets such as GSE84589 containing purified rods and cones from Mus musculus were used to validate rod and cone gene lists through differential expression. B Rod cell type-specificity of rod gene lists from 4 methods: Boolean implication, SRCCA, SRCCA filtered using Boolean implication and SRCCA without Boolean implication. C Photoreceptor-specificity of rod gene lists from 4 methods. D Cone cell type-specificity of rod gene lists from 4 methods. E Validation of cone and rod genes from Boolean analysis of large single-cell dataset GSE63472. Validation of the genes’ specificity was assessed in bulk dataset GSE84589. “Cones vs. Rods” denotes the specificity of putative cone genes in comparison with rods, and vice versa. “Vs. Retina” denotes the specificity of putative cone/rod gene in comparison with the entire retina. Abbreviations: Corr., correlation; Bool., Boolean; SRCCA, Spearman’s rank correlation coefficient analysis. Note: P values are from two-proportion Z-test between proportion of cell type-specific genes in lists from SRCCA and SRCCA filtered using Boolean implication