to 1%, while all other parameters were set as recommended for Roche/454 Titanium. 85.4% of these reads aligned. Simulated reads were drawn from a set of B. aphidicola haplotypes, created by GemHaps using the B. aphidicola Cc reference genome [GenBank ID: PC000263.1]. The number of simulated reads was five million and one million for Illumina GAII and Roche/454 FLX Titanium respectively. This gave a reference coverage of around 1000×. For consistency, alignment of simulated reads was done using Novocraft for the Illumina data and Mosaik for the Roche/454 data [18]. Memory and runtime data are summarised in Table 1 and show that GemSIM can be run with modest memory requirements (< 1 Gb) on a single CPU/desktop computer within a reasonable time frame. As memory requirements are dictated mainly by the size of the sequence-context word and read length, they are largely independent of the number of reads processed or simulated. Runtime scales linearly with the total number of bases processed or simulated.
