Fig. 4: Isoform quantification recapitulates 3’ UTR shortening during mouse spermatogenesis and identifies novel cell types.

a Cell types identified using gene expression (left) and isoform expression estimated by SCALPEL (right). Both analyses identify spermatocytes (SC), round spermatids (RS), and elongated spermatids (ES). Agreement in clustering solutions is shown by a Sankey plot. b Scatter plot showing the cell-type expression specificity relative to gene expression for all genes with only two expressed isoforms. The higher the information content of a gene, the more cell-type-specific its expression is. Colored dots represent genes whose isoform expression usage changes across conditions (Chi-squared test adjusted p-value < 0.05). c SCALPEL quantification of isoform usage from Ighmbp2. Coverage plots show the distribution of filtered reads in SC, RS and ES clusters. SCALPEL quantifications of the different isoforms are shown on top of the custom tracks. d Number of DIU genes identified in each pair of samples that are also differentially expressed (light gray) or not (dark gray). Most genes show changes in isoform usage independently of changes in expression. e Classification of DIU genes depending on the type of APA event identified as alternative last exons (ALE), tandem APA sites (tandem APA), and others. f Quantification of Smg7 isoforms by SCALPEL. Coverage plots show a gradual switch in isoform usage during the differentiation from SC to ES cells. g Pseudotemporal ordering of cells confirms the overall shortening of 3’ UTRs during mouse sperm cell differentiation. h High-resolution clustering using isoform expression data (right) identifies novel cell states (RS6) that cannot be identified using gene expression data (left). Using this resolution, we identified 15 cell populations in the isoform-based analysis including spermatocytes (SC1-5), Round cells (RS1-6), Elongated Spermatids (ES1-2), and Condensing Spermatids (CS1-2). The new RS6 population identified in the isoform-based analysis is composed of cells coming from several RS populations identified using gene expression data. i GO term enrichment analysis using RS6 marker isoforms identified significant terms associated with cilium organization and cell differentiation. In the circular barplot, GO terms are arranged according to their combined enrich R score and colored according to their adjusted p-value. Source data are provided as a Source Data file.