HiSeq data sets 10 to 14 and A1 to A3; and 2 × 250 bases for HiSeq 2500 data sets A10 to A11. Data were further processed using the Picard data-processing pipeline [42] to generate BAM files. Alignment was performed using BWA version 0.5.9 [43]. The 'aln' command was run with the alignment options '-q 5 -l 32 -k 2 -o 1', followed by the 'sampe' command to generate a paired-end alignment. The Picard MarkDuplicates program was applied after alignment and all duplicate-flagged reads were excluded from the analyses in this manuscript. All human data sets, with the exception of data set 10 to 12, were also processed with the GATK IndelRealigner and TableRecalibration tools [44,45], but none of the results presented in this work depend on precise indel placement or on quality scores.
