The model and inference algorithm is implemented in C and supports both SAMtools (Li et al., 2009) and Maq pileup format. Running EM (SNVMix2) on 14 649 positions for the 40× breast cancer genome took 36 s. Predicting genotypes for the whole 40× genome took 11 min and 38 s. (The Maq step cns2snp took 19 min and 9 s.) A script to choose optimal base and mapping quality thresholds, given ground truth [or orthogonal single nucleotide polymorphism (SNP) array] data, is provided in the software package.
