Skip to main content
. 2020 Aug 24;9:e55792. doi: 10.7554/eLife.55792

Figure 5. The V4.2 annotation increases the quantity of cell type-specific genes identified by scRNA-seq.

(A) Venn diagrams illustrating intersection by common LL gene ID of genes enriched in mia-positive cartilage cells and and1-positive epidermis cells by both indicated annotations. (B) 3' UTR lengths from Ens95 and V4.2 for reference genes identified as cartilage or epidermis-specific uniquely by V4.2. Data are not normally distributed (Shapiro-Wilks test), comparison by Wilcoxon matched-pairs signed-rank test, p-values are shown. (C) Volcano plots showing cartilage- or epidermis-specific genes identified selectively by V4.2 with corresponding values from bulk RNA-seq comparison of pdgfrbpos and pdgfrbneg cells. Red indicates log2 fold change >1 (pdgfrbpos/pdgfrbneg), padj <0.05. (D, E) tSNE plots of the mia-positive cartilage cluster showing expression levels of sox9a and fzd9b using scRNA-seq data quantified using (D) Ens95 and (E) V4.2. Legends indicate log-transformed and normalized expression levels per cell. (F) UCSC browser image of sox9a and fzd9b loci showing 3' UTR annotations from V4.2 and Ens95. Both loci are located on the negative strand. Mapped read depth from kdrlpos cells on the genome, or assigned to each annotation are indicated, as are 3P-seq features, all of which reside in the same orientation as the genes themselves.

Figure 5—source data 1. Cartilage-specific genes identified by scRNA-seq.
Worksheet one includes all cartilage genes identified using both Ens95 and V4.2 quantification, as indicated, with associated adjusted p-values and log2 fold change (comparison of cartilage cells to all other clusters). 3' UTR length from matching reference gene (from Figure 2—source data 2) is indicated. Worksheet two includes matched bulk RNA-seq quantification from Ens95 and V4.3 annotations for cartilage-specific genes. Data used to generate Figure 5A–C, Figure 5—figure supplement 1A,B.
Figure 5—source data 2. Epidermis-specific genes identified by scRNA-seq.
Worksheet one includes all epidermis genes identified using both Ens95 and V4.2 quantification, as indicated, with associated adjusted p-values and log2 fold change. 3' UTR length from matching reference gene (from Figure 2—source data 2) is indicated. Worksheet two includes matched bulk RNA-seq quantification from Ens95 and V4.3 annotations for epidermis-specific genes. Data used to generate Figure 5A–C, Figure 5—figure supplement 1A,B.

Figure 5.

Figure 5—figure supplement 1. Improved detection of cartilage and epidermis genes in scRNA-seq data using V4.2.

Figure 5—figure supplement 1.

(A, B) Left, plots showing values from bulk RNA-seq comparison of pdgfrbpos and pdgfrbneg cells quantified with Ens95 for (A) cartilage and (B) epidermis genes identified as such selectively in V4.2 by scRNA-seq. Red indicates log2 fold change >1, padj <0.05. (A, B) Venn diagrams showing the intersection of (A) cartilage and (B) epidermis genes identified as pdgfrbpos-enriched in bulk RNA-seq using indicated annotation. (C) Examples of epidermis genes identified by scRNA-seq using V4.2, but not Ens95. Left, tSNE plots of epidermis cluster from data quantified with Ens95 or V4.2 showing expression of fbln1, mmp11a, and ptx3a. Right, 3' UTR annotation for each gene from Ens95 or V4.2. Location and read density of scRNA-seq reads are indicated.