While initially measured at the base level, over-call/under-call errors are the consequence of inaccuracy at the flow-value level. Thus, relationships between error occurrence and features of the read and sequencing parameters are best examined directly at the flow-value level [15]. Inspection of the error rate by flow position also shows a positive correlation between the homopolymer error rate and the flow position (Figure S9). However, the error rates per flow are substantially higher than those observed per base, with homopolymer error rates approaching 20–30% at flow position 350 onwards for the 200 bp One Touch kit. Surprisingly, there is also an observed periodicity in the error rate, with the period remaining approximately constant, however the amplitude increases with distance from the first flow. Intuitively, this periodicity suggests that the error rate may be influenced by the position-within-a-flow cycle (PIC) in a non-linear way. Inspection of the relationship between PIC and error rate suggested no obvious parameterised function, so each PIC (0–31) is considered a different ‘factor’.
