RNA-seq transcript quantification from reduced-representation data in recount2

Abstract

More than 70,000 short-read RNA-sequencing samples are publicly available through the recount2 project, a curated database of summary coverage data. However, no current methods can be directly applied to the reduced-representation information stored in this database to estimate transcript-level abundances. Here we present a linear model taking as input summary coverage of junctions and subdivided exons to output estimated abundances and associated uncertainty. We evaluate the performance of our model on simulated and real data, and provide a procedure to construct confidence intervals for estimates.