Fig. 4: Emergence of non-coding transcript biotypes is a general feature of terminal differentiation.

A–D Analyses of scRNA-seq data on mouse lung alveolar type II (AT2) cell development. E–H Analyses of scRNA-seq data on in vitro differentiation of human neural precursor cells (NPCs) to neurons. A, E Monocle-generated trajectory plots representing differentiation pseudotime ordering of mouse AT2 cells (A) and human neurons (E). E, embryonic day; P, postnatal day; D, day. B, F Scaled and centered expression heatmaps of transcripts with dynamic expression over the mouse AT2 cells (B) and human neurons (F) differentiation pseudotimes. Transcripts are ordered by row and cells are by column. Transcripts are clustered by their expression patterns over pseudotime. Number of transcripts per cluster is shown. C, G Bar charts showing the relative proportions of transcript biotypes in each pseudotime-dependent transcript clusters and in all cells (C, mouse AT2 cells; G, human neurons). All others: 11,052 all other detected transcripts with TPM > 10 in more than 9 AT2 cells (C), and 22,291 all other detected transcripts with TPM > 10 in more than 49 neurons. Chi-squared test: p < 2.2 × 10⁻¹⁶ for protein-coding vs. non-coding transcript numbers across all clusters, including “All others”. D, H Top gene ontology terms associated with genes encoding non-coding transcript biotypes in transcript clusters 4 and 5 (highest expression in late/differentiated mouse AT2 cells (D) and human neurons (H)). Bar magnitude, enrichment score; dot, p-value. See also Supplementary Figs. 9–13.