Fig. 3

Recall achieved by the different methods vs the expression level of true transcripts in the two 40-million read-pair data sets. Transcripts with simulated FPKM >0.1 (total) are grouped into sets low, high, and med that contain the lowest expressed 20 % of the transcripts, the highest expressed 5 %, and all remaining transcripts, respectively. a 75-bp data set. b 100-bp data set