For the 100 bp OneTouch kit and 200 bp manual kit, the great majority of reads (94–100%) had adapter sequences detected by the PGM software. The newest kit considered here, the 200 bp OneTouch Kit, had a very low percentage of reads with detected adapters (7–12%), which may be due to library construction differences (longer inserts used). We applied the 3′ clipping of adapter sequences as specified in the Standard Flowgram File (SFF), however we did not apply the recommended 3′ quality clipping – allowing more realistic calculation of raw error rates. Datasets larger than 300,000 reads were randomly subsampled down to 300,000 reads as this provided sufficient information for downstream analyses ( Table 1 ).
