Question: Does data analyzed with intronic reads have biases as compared to data analyzed without intronic read counting?
Answer: Data analyzed with intronic reads has biases different from the data analyzed without intronic reads. The biases can be attributed to the underlying source of the intronic reads. Intronic reads potentially arise from priming off of intragenic poly-A tracts. The relative increase in UMI counts when including intronic UMIs has a gene length bias, possibly because longer genes have more poly-A tracts. There is also a potential bias for poly-A rich genes. Further, the data indicates that transcripts from long genes may be assigned multiple UMIs due to priming at multiple sites. More details can be found in the Technical Note here.
Despite above biases between the two modes, our data indicates that meaningful gains in sensitivity are made by using intronic reads. These gains lead us to believe that intronic reads should be counted for whole transcriptome gene expression analysis.