A small number of junctions predicting very small and very long introns may be mispredictions and should be interpreted with care; these are currently being re-examined.
For ~1900 predicted junctions, the supporting evidence came from a methodology (Wang L, Xi Y, Yu J, Dong L, Yen L, et al. 2010 A Statistical Method for the Detection of Alternative Splicing Using RNA-Seq. PLoS ONE 5(1): e8529. doi:10
.1371/journal.pone.0008529) that did not report number of supporting junction-spanning reads. Each of these ~1900 predicted junctions was supported by at least two reads, and so we have noted these by having a total junction-spanning read count of ">1". However, we do not know which developmental stages these supporting reads were in, and so all developmental stages are reported as "0".