paperKB
coga / coga-kb
Help
Sign in

Chunk #25 — Results — Homopolymer errors

Source
Shining a light on dark sequencing: characterising errors in Ion Torrent PGM data.
Embedded
yes

Text

As with Roche 454 [11], [14], [15], base-calling accuracy decreases with the length of the homopolymer, with 6-mer calling accuracy approaching 68%±6.9% s.d using the OneTouch 200 bp kit ( Figure 5 ). However, unlike Roche 454, insertions are more likely than deletions when the ‘homopolymer’ length is less than two (Figure S8). In fact, the accuracy of calling a zero length homopolymer was between 98.75% and 99.72% across the kits (a substantial contributor to the overall high insertion-rate). While insertions are the dominant error-type, they are more common for homopolymer lengths less than two, but are rapidly overtaken by deletions for homopolymers of length two or more (Figure S8). The accuracy in calling a homopolymer is highly variable, even for short homopolymers (s.d of 0.5%–1.7% for 2-mer accuracy) suggesting that base-calling accuracy is not dependent on homopolymer length alone.