Fig. 1

Overview of the MicroExonator workflow. a To discover unannotated microexons, RNA-seq reads are aligned with BWA-MEM to the annotated splice junctions. The resulting alignments are postprocessed to discover novel microexons flanked by canonical U2-type GT-AG splice sites. b Both putative novel and annotated microexons are quantified and filtered to produce a final list of microexon incorporation into transcript models which can be used for downstream analysis. c Number of intronic matches and distribution of spurious match probability across microexon lengths. d A two-component Gaussian mixture is used to fit the U2 consensus splicing score distribution. The red line shows the Gaussian with lower mean U2 splice score, which is assumed to consist mainly of false positives, while the green line denotes the Gaussian with higher mean U2 splice-site score