Simulation of NGS data, followed by software benchmarking, presents an alternative approach. Early published simulators include GenFrag [5]. Its usefulness for modern NGS projects is limited by a simplistic error model, a single input genome, and a lack of quality score information. SamTools [6] also supplies a simulator, however it uses a uniform error rate. A uniformly increasing error rate is used in a slight improvement released as 'dwgsim'[7]. NGS features highly heterogenous error profiles [8,9], so the usefulness of this simulator must be questioned.
