A novel splicing graph allows a direct comparison between exon-based and splice junction-based approaches to alternative splicing detection

There are primarily two computational approaches to alternative splicing detection: splice junction-based and exon-based approaches. Despite their shared goal of addressing the same biological problem, these approaches have not been reconciled before. We devised a novel graph structure and algorithm aimed at mapping between the exonic parts and splicing events detected by the two different methods. Through simulations, we demonstrated disparities in sensitivity and specificity between splice junction-based and exon-based methods. When applied to empirical data, there were large discrepancies in the results, suggesting that the methods are complementary. With the discrepancies localized to individual events and exonic parts, we were able to gain insights into the strengths and weaknesses inherent in each approach. Finally, we integrated the results to generate a comprehensive list of both common and unique alternative splicing events detected by both methodologies. Availabilityhttps://github.com/HanLabUNLV/GrASE Contactmira.han@unlv.edu Supplementary informationSupplementary data are available online.