(A) The previously identified 5’ splice sites (SS) and 3’ SS extracted from reference (BDGP v5.78) were used 206,299 and 207,082 times in the tsu sample, respectively, while those used in the lacZ sample were 204,053 and 204,713 times. Thus, compared with the lacZ sample, the tsu sample has a 1.4% (2246/204,053) higher usage for previously identified 5’SS and a 1.5% (2269/204,713) for 3’SS, respectively. p-value for 5' splice site usage is 0.079, and 0.057 for 3’ splice sites (Student’s t-test). (B) Previously unidentified splice sites were detected in RNA-seq when the pre-EJC was knocked down. In total, 2,207 of 5’ SS and 2,081 of 3’ SS were identified, among which 394 of 5’ SS and 395 of 3’ SS were located in annotated exons in reference (BDGP v5.78), respectively. (C) Distribution plot highlights the correlation between maximum intron length and genes whose splicing were subject to differential regulation by the pre-EJC. Shown are distribution plots of genes whose splicing pattern was neither affected (black line) nor changed (red line) by reduced pre-EJC activity. The genes with altered splicing have an overall larger maximum intron length (average length>1000 nt). This result is consistent with previous reports (Ashton-Beaucage et al., 2010; Roignant and Treisman, 2010). In addition, a new class of genes was identified in which previously unidentified splice sites were utilized for generating novel transcripts (green line). Interestingly, no correlation with overall larger maximum intron length was detected in this new class of genes (green lines), implying an unknown mechanism for the pre-EJC to recognize splice sites.