Fig. 1

Profiling APA from RNA-seq. a Overview of annotated 3′ UTR library generation and QAPA method. Top: Terminal exons of two alternative 3′ UTR isoforms. The grey box indicates the coding sequence region. The blue region indicates the common region shared by both isoforms. The green region indicates the alternative region found only in the longer isoform. In (1), additional poly(A) site annotations (inverted chevrons) are used to refine the 3′ coordinates, as well as establish new isoforms. These new sequences are then used in (2) to measure expression from RNA-seq data and in (3) to estimate relative alternative 3′ UTR isoform abundance. b Hexbin scatterplot comparing PPAU estimates of 975 genes derived from HEK293 control samples assayed by RNA-seq (QAPA) [34] and A-seq2 [14]. Bins are colored by number of data points and the dashed line indicates the reference diagonal. c Scatterplot comparing ∆PPAU for 86 highly expressed genes between human skeletal muscle and brain tissue samples from RNA-seq (QAPA) [35] and 3′-seq [20]. d Receiver operating characteristic curves comparing performance of QAPA and other methods on simulated RNA-seq data. e Bar plot showing average runtime of each method on the same four RNA-seq samples divided into “pre-processing” stage for method-specific data preparation and “APA” stage for direct computation of APA results