Error correction using SA was applied to each set of reads after trimming ends of reads with phred scores averaging <20 over five bases. Reads were mapped to the original genome by querying a compressed suffix array as described in Lippert et al. (2005).