Defining Loci in Restriction-Based Reduced Representation Genomic Data from Nonmodel Species: Sources of Bias and Diagnostics for Optimal Clustering
Figure 3
Clustering mismatch threshold series for simulated stickleback reads (a), simulated soybean reads (b), simulated C. savignyi reads (c), and experimental C. savignyi data (d). The -axis represents the percentage of total clusters for a given organism at a given mismatch value and the -axis represents the maximum proportion of differences (mismatches) allowed between reads within a cluster. Single haplotype clusters (putative homozygous loci) are represented by a solid blue line and diamonds, two-haplotype clusters (putative heterozygous loci) are represented by a solid green line and squares, and three or more haplotype clusters (combined alleles from 2 or more paralogous loci) are represented by a solid red line and triangles. Striped shaded areas for simulated data represent deviation from the true values due to assembly artifacts such as splitting alleles into different clusters when the threshold is low or combining paralogs into a cluster when the threshold is high. Solid shaded areas for experimental data represent deviation from the expected simulated values due to assembly artifacts and null alleles. The uptick in heterozygosity observed between 0.8 and 0.10 in (b) is likely a result of the surge in 2+ paralog clustering over the same interval, perhaps due to clustering of duplicated loci from the soybean polyploidy event. The mean values of total cluster counts across assembly thresholds for plots (a)–(d) are 88300 (SD = 1987), 29201 (SD = 4798), 7700 (SD = 1244), and 167824 (SD = 10863).
