Fig. 4

Effect of sequencing depth on the numbers and correctness of transcripts assembled from long-reads data. A. Transcript counts from iPSC and H9 human cell lines. 25% (n = 4), 50% (n = 2) and 75% of the H9 reads were independently sampled without replacements. B. Numbers of assembled transcripts from down-sampled H9 long-read data. Effects of down-sampling on coding to noncoding ratio (panel C), sensitivity (panel D) and precision (panel E). The transcript assembly from all the H9 long-read data was used as the reference for the computations of sensitivity and precision. F. The proportions of TE-containing transcripts across various depths. The sensitivity of the assemblies with or without TEs in coding (G) and noncoding (H) transcript sets. I. The cumulative expression of transcripts ranked by expression levels in H9 samples for 25% and 75% subsampled read assemblies. J. The cumulative precision of the expression-ranked transcripts for 25% and 75% subsampled read assemblies. K. The cumulative coding to noncoding ratio of the expression-ranked transcripts