Fig. 2
From: Using expected sequence features to improve basecalling accuracy of amplicon pyrosequencing data
Ranking of the correct basecall according to P(CBS|flows). Upon flowgram clustering and alignment, Multipass was employed to calculate the fifty most likely basecalls given each flowgram alignment. The likelihood ranking of the correct basecall is shown against the number of flowgrams in the alignment. Maximally two hundred flowgrams from each cluster were used for alignment. The marker size is proportional to the abundance at a point