Figure 1

Workflow of Rare Variant Detection (RVD) Application. The standalone application requires four inputs: a configuration file containing directory locations, base quality and resolution thresholds, and region of interest information, a metadata information file with sample-specific information, a reference sequence file, and BAM format files with short read sequencing data for each sample. A beta-binomial prior distribution is applied to estimate the error rate of reference replicates that are sequenced alongside samples. In the figure, θ is the binomial parameter that represents the true error rate at a single position in a single experimental replicate. The variable θ has a beta distribution with parameters {M_0, μ} where μ is the prior error rate estimated at the given position and M_0 is the experimental precision of the error rate. The output of the application is a tab-delimited text file containing, for each position in the reference sequence, an estimate of the minor allele frequency (MAF) in the sample and a statistical call for significance. RVD is able to detect single nucleotide variants with minor allele frequency as low as 0.1%.