Fig. 3

De novo transcript assembly of healthy and dysplastic HSCs. a Number of transcripts overlapping repeats in GENCODE v29 (comprehensive gene annotation) and the de novo assembled transcriptome of healthy, MDS, CMML and AML HSCs. b Representation of multiexonic or monoexonic repeat-overlapping transcripts, according to the transcript structure in GENCODE v29 and the de novo assembled transcriptome. c De novo assembled transcripts that are differentially expressed between healthy and untreated dysplastic HSCs were identified by multigroup comparison (q ≤ 0.05). Heatmap of the expression of all the identified transcripts and hierarchical clustering of samples according to this expression. HSCs from all untreated patients from both cohorts are shown. d Hierarchically clustered heatmap of 5-AZA-responsive de novo-assembled transcripts in MDS and CMML bone marrow HSCs in our study (left) and in SRP067631 (right) (≥ 2-fold, p ≤ 0.006). Only patients from the second cohort of our study are shown