Isoform prefiltering improves performance of count-based methods for analysis of differential transcript usage

Count-based methods generally perform well in the detection of differential transcript usage. Controlling the false discovery rate at the imposed threshold is difficult, particularly in complex organisms, but can be improved by prefiltering the annotation catalog.