Fig. 4

Effects of ploidy and heterozygosity levels on accuracy and contiguity. Through these graphs, we show the effects of sample properties (heterozygosity level and ploidy) on nPhase’s accuracy metrics when run with default parameters. a Each bar displays the contiguity of a different test result. The least contiguous heterozygosity level is 0.05%, likely related to its also yielding the least accurate results. Overall, we note little absolute variation in the contiguity. Interestingly, contiguity at higher heterozygosity levels appears to be a function of ploidy. Higher ploidies seem less likely to become less contiguous as a result of increasing the heterozygosity level, while the diploid tests are more affected. We also note that tetraploids of high heterozygosity level are the most contiguous. b Each bar displays the accuracy of a different test result. As ploidy increases, the accuracy tends to decrease. It also appears to decrease faster for tests on low heterozygosity level constructions. c, d Each bar displays our evaluation of the effects of ploidy and heterozygosity level on the error and missing rates, respectively, for our 12 tests using optimal parameters. Overall, we see that the error rate is always higher than the missing rate across these conditions. As the heterozygosity level increases, the error and missing rates decrease along with the gap between ploidies. We also find that more difficult phasing problems (high ploidy and low heterozygosity level) yield much higher error and missing rates and that the low heterozygosity tetraploids seem to be particularly sensitive to missing calls