Fig. 2

Improvement in Clair SNV calls from nanopore data. a Base quality and mutation frequencies obtained for 1 million randomly selected 5-mers with SNV in the base 3 position (1:1 true positives to false positives). b PCA analysis of the reference 5-mer GTACT shows separation of true positive from false positives based on quality scores and mismatch frequencies. c Clair variant calling quality distribution for the NA19240 run 0 sample. d Quality distribution upon normalization of Clair’s qualities using the weights given by SNVoter to each SNV. The highlighted region represents the optimal threshold area to filter out low-quality calls. e Receiver operating characteristic curves for SNV calling using Clair or using Clair+SNVoter for different coverage depths. NA19240 run 1, NA19240 run 0, and Colo829BL are processed by SNVoter using the model trained on NA12878 20FCs (24×). NA19240 runs 0&1 and NA19240 runs 1&2 is processed using the model trained on NA12878 whole dataset (44×)