Performance of PAINTOR compared to standard methodologies at variable sized loci.

To expedite simulations, we used a modified version of the simulation setup. As before, causal SNPs were drawn according to a logistic prior such that in expectation there were a total of 100 causal variants – we did not enrich causal in any annotations. For this experiment, Z-scores were drawn directly from a multivariate normal distribution; this gave virtually identical results to using simulated genotypes derived from HAPGEN (see Methods). We find that PAINTOR increasingly outperforms existing methodologies as the size of the loci become larger.Performance of PAINTOR compared to standard methodologies at variable sized loci.为提升模拟效率，我们采用了经修改的模拟实验配置。 与前期实验一致，我们基于逻辑先验分布抽取因果SNP（causal SNP），使得预期总因果变异位点数量为100；且未针对任何注释类别富集因果变异。 本次实验中，我们直接从多元正态分布中抽取Z分数（Z-score）；该方案得到的结果与使用HAPGEN模拟得到的基因型结果几乎完全一致（详见方法部分）。 我们发现，随着基因座（locus）规模的扩大，PAINTOR的性能优势会愈发显著，逐步超越现有方法。 不同规模基因座下PAINTOR与标准方法的性能对比。