Alignment and mapping methodology influence transcript abundance estimation

We observe that, on experimental datasets, the performance of lightweight mapping and alignment-based approaches varies significantly, and highlight some of the underlying factors. We show this variation both in terms of quantification and downstream differential expression analysis. In all comparisons, we also show the improved performance of our proposed selective alignment method and suggest best practices for performing RNA-seq quantification.