Figure 2 Simulation Results Using Flux Simulator, we ran six simulations with varying levels of unannotated transcripts. Subsequently, we ran quantifications with three methods with the known GENCODE v.12 annotation. We compared the simulated versus measured link quantifications via Spearman’s rank correlation. These comparisons are shown as colored solid lines. In order to produce a null random distribution for each method, we took the link quantifications for each gene, permutated these for 100 times within the links of this gene, and measured the correlation of these random assignments with the simulated ones. By using this sampling method stratified by genes, we account for the variability of number of isoforms per gene. These correlations for random assignments are shown as dashed lines. We observe that as the percentage of novel transcripts increase, the performance of Cufflinks and MISO suffer, whereas this is not the case for Altrans, which results in best quantifications with increased levels of unannotated transcripts.