Skip to main content
Fig. 3 | BMC Bioinformatics

Fig. 3

From: Using expected sequence features to improve basecalling accuracy of amplicon pyrosequencing data

Fig. 3

Receiver operator characteristics for prediction of incorrect sequences. Performance of three machine learning methods applied to differentiate between sequences with and without error in var sequences generated using Multipass basecalling. The classifiers were trained on a number of characteristics provided for each sequence, such as read coverage and maximal positional flow variance. Positive (P) and negative (N) refers to sequences with and without error, respectively. True (T) and false (F) refers to correct and incorrect predictions, respectively. For each method, the lowest false positive rate with perfect classification of the erroneous sequences is indicated (dotted lines)

Back to article page