SNP predictions were generated using a combination of SSAHA2 (Ning et al. 2001) to map the reads, Samtools (Li et al. 2009) to generate the variant predictions, and in-house Perl scripts to further filter the results. SNP-dense regions (more than three SNPs in a 7-bp window) were excluded, as were SNPs within 100 bp of contig edges. SNPs with consensus quality and base quality scores > = 40, mapping quality scores > = 25, coverage > = 10 reads, and less than twice the median coverage of the chromosome were retained. Finally, SNPs in the bin contigs of L. infantum, L. mexicana, and L. braziliensis were excluded due to uncertainty of median read coverage for these regions.