In this study, we considered the influence of both technological (template preparation kit, chip, machine) and DNA template factors on sequencing output using a full factor statistical design (see Methods ). We evaluated three template preparation kits, listed in chronological release order, Ion OneTouch Template Kit (100 bp OneTouch kit), Ion Xpress Template 200 Kit (200 bp manual kit), Ion OneTouch 200 Template Kit (200 bp OneTouch kit), two chip densities, 314 (1.2M sensors, density 200,000 reads) and 316 (6.1M sensors, density 1 million reads) and two PGM machines (named here a and b). Two species, Sulfolobus tokodaii (33 G+C%) and Bacillus amyloliquefaciens (46% G+C%), both with relatively small, sequenced genomes and no extra-chromosomal DNA, were selected to represent different G+C% templates. A third organism, Deinococcus maricopensis (69% G+C), was selected for its high G+C content, however library preparation consistently failed using the 200 bp kits, and read throughput was extremely low for the 100 bp OneTouch kit, consequently it was excluded from subsequent analyses. Fifteen datasets were generated ( Table 1 ) and used to evaluate the effects and interactions between template preparation kit, chip and machine on the measured error rate (see Methods ).
