Fig. 2

Impact of mismatches around splice junctions on the accuracy of their determination. A Splice junctions (SJs) shared by AtRTD2 and Iso-seq (LDE_30; sjt_30) and unique to each. B Position weight matrix (PWM) scores for splice sites unique to Iso-seq transcripts and shared with AtRTD2. PWM scores for 5′ and 3′ splice site sequences from SJs shared between AtRTD2 and Iso-seq transcripts (high confidence), are significantly higher (t-test, p < 2.26e−16) than those unique to Iso-seq (low confidence). C, D Distribution of the number of mismatches in each position 30 nt upstream (C) and 30 nt downstream (D) of SJs unique to Iso-seq (low confidence) and shared with AtRTD2 (high confidence). See Additional File 1: Tables S3A,B). E Filtering of SJs—the graph shows the number of SJs remaining (expressed as a percentage) after the cumulative removal of SJs with mismatches in the first n positions (1, 2, 3, etc.) flanking SJs. See Additional File 1: Tables S5A,B)