Both PCR and gel extraction processes have been implicated in G+C% coverage biases in Illumina sequencing [6], [7]. PCR has been shown to preferentially amplify mid-G+C% fragments, and size selection using gel extraction approaches, which heat the DNA, often leading to underrepresentation of low G+C% sequences. In this study, size selection was performed using the Pippin Prep (see Methods ) circumventing gel extraction, leaving the PCR steps as the likely cause of the coverage biases observed in low and high G+C% regions. As PCR biases are exerted only on sequences processed together, it is expected that the observed relationship between G+C% and coverage will depend on the mean and range of G+C% in the DNA template. As demonstrated here, organisms with 100 bp G+C% windows within the 20%–80% G+C range will experience little bias in coverage, whereas organisms with G+C% regions outside this ‘safe’ range will suffer substantial G+C biases. The G+C% bias appears asymmetric, with the bias against high G+C% regions (i.e. D. maricopensis) more pronounced than low G+C% regions (i.e. S. tokodaii). In fact this bias against extremely
