In Table 2 we can also see how updates to the Illumina HiSeq platform have affected bias. Notably, the HiSeq version 3 data (data sets 13 and 14) show better coverage of high-GC motifs and the bad promoters compared to the HiSeq version 2 data (data sets 10 to 12). We have also compared the standard list of bad promoters, computed from HiSeq version 2 data, to a new list computed from HiSeq version 3 data (see Materials and methods and Additional files 1 and 2 for details). The lists have 47% of their bases in common, which indicates that many bad promoters are still resistant to sequencing despite Illumina's improvements.
