Figure 3

The effect of different assembly errors or structural variation on clustering. Different errors will generate characteristic patterns in the clustering data. Consider two scaffolds in close proximity on a chromosome, scaffold_1 and scaffold_2. (A) In a case where both scaffolds are oriented in the same direction, the scaffolds will have the same strand-state patterns. When comparing homozygous patterns (WW scaffolds against CC scaffolds), heterozygous patterns (WW or CC scaffolds against WC scaffolds) or comparing all three strand states against each other, there will be high similarity. (B) In the case of a misorientation (or a homozygous inversion), the strand-state patterns will be antithetical when comparing homozygous states, as whenever scaffold_1 is WW, scaffold_2 will be CC, and as such, these scaffolds will be completely dissimilar. However, since misorientations are not visualized in heterozygous inheritance patterns, when comparing WW or CC states against WC states, the scaffolds are highly similar. When comparing all three states against each other, the similarity seen with WC scaffolds and dissimilarity seen with WW or CC scaffolds will cancel out, resulting in ~50% similarity. (C) In cases of a heterozygous inversion, either scaffold_1 or scaffold_2 may have a homozygous state, but not both. Therefore, no comparisons can be made when only considering the homozygous states, and NA values are generated. There will, however, be a high degree of dissimilarity when comparing homozygous and heterozygous states. It is important to distinguish these natural structural variants from assembly reference errors. (D) In cases where a scaffold is incorrectly located to a chromosome (i.e., a chimera), the inheritance pattern between the two scaffolds will be random, and there will be no significant similarity or dissimilarity between these scaffolds.