Relationship between CDS consistency and the sequence coverage.

Shotgun reads in library 1 and 1+2 were randomly sampled to simulate datasets of different sequence coverage. Each simulated dataset was then assembled using MuSICA 2, and the CDS structure consistency for the 199 reference full-length cDNA clones was evaluated. Full data corresponds to 100% sequence coverage. The definition of CDS structure consistency is exactly same as in Table 2. Note that the purity filtered reads, not the raw reads, were used for input to MuSICA 2, which accounts for a slight difference between Table 2 and the number of CDS-consistent/inconsistent clones at 100% sequence coverage.