Fig. 1

Sensitivity in recovering the variants in the truth set using the Leucegene RNA-Seq samples. a Median with maximum and minimum sensitivity (vertical bars) for recovering the SNVs (left plot) and indels (right plot) in Table 1, across random downsampling runs using different library sizes. Each estimated median sensitivity represents the median across 5 random downsamplings (only 3 for 80M libraries) of the initial RNA-Seq libraries at a specific library size. The solid lines are the sensitivities obtained using the default-filters and the dotted lines are obtained with the annotation-filters. b Average VAF (top plot) and alternative depth (bottom plot) on the log scale at a variant site for the variants in the truth set using different library sizes. A line in each plot represents one mutation in the truth set. Each dot is coloured according to the average number of times a variant was called by one caller using the annotation-filters across replicated downsampling runs at one specific library size. c Heatmap showing the sensitivity in detecting the Leucegene mutations using default-filters across intervals of the VAF at a variant site and intervals of the total log2 gene counts of a gene across different library sizes. Every coloured square in the heatmap represents the average time the Leucegene variants within an interval were detected. The average is over mutations found within an interval and over callers