Additional file 1 of Modeling the longitudinal changes of ancestry diversity in the Million Veteran Program

Additional file 1: Table S1. Patterns of service era per birth cohort and across all MVP participants stratified by sex and HARE superpopulations. Each row represents a distinct pattern of service across nine service eras; the frequency of each is calculated by birth cohort and for all MVP participants. Service patterns with less than 11 participants were omitted to preserve data privacy of the participant so HARE total population sample sizes are slightly lower than those reported in Table 1. Table S2. Sample size per birth cohort derived from cumulative distribution function of year of birth. Table S3. Mean ancestry proportion of five 1kGP reference populations in all birth cohorts and HARE superpopulations. Two-sided Z-tests were used to compare the statistical difference in means between groups and the corresponding p values reflect this difference. Standardized mean differences reflect the magnitude of effect size difference between two groups. Table S4. Comparison of height across birth cohorts in each MVP HARE superpopulations. Table S5. Metrics for GWAS of height in each ancestry per birth cohort using both methods of population assignment. Heritability, LDSC intercepts, and attenuation ratios were compared across birth cohorts, within each method, using two-sided Z-tests. Multiple testing correction was applied using a false discovery rate of 5%; differences surviving multiple testing correction are highlighted in yellow. Table S6. Metrics for GWAS of height compared across method used to define superpopulations. Two-sided Z-tests were used to compare heritability, LDSC intercepts, and attenuation ratios between HARE and 1kGP+HGDP superpopulation assignments. Multiple testing correction was applied using a false discovery rate of 5%.

附加文件1：表S1。按性别与HARE超级种群（HARE superpopulations）分层的所有MVP参与者（MVP participants）以及各出生队列的服务时代模式。每行代表9个服务时代中一种独特的服务模式；每种模式的出现频率分别按出生队列及全部MVP参与者计算。为保护参与者的数据隐私，参与者人数少于11人的服务模式均被剔除，因此HARE总体人群的样本量略低于表1中报道的数值。
表S2：基于出生年份累积分布函数得到的各出生队列样本量。
表S3：所有出生队列及HARE超级种群中，5个1kGP参考人群（1kGP reference populations）的平均祖先占比。采用双侧Z检验比较组间均值的统计学差异，对应的P值反映该差异的显著性；标准化均值差则反映两组间效应量差异的大小。
表S4：各HARE超级种群中的MVP参与者群体内，不同出生队列的身高水平对比。
表S5：基于两种人群划分方法，各出生队列中不同祖先群体的身高全基因组关联研究（GWAS）分析指标。针对每种分析方法，采用双侧Z检验比较不同出生队列间的遗传力、LDSC截距（LDSC intercepts）及衰减比；采用5%的错误发现率进行多重检验校正，经该校正后仍显著的差异以黄色高亮标注。
表S6：基于不同超级种群划分方法的身高全基因组关联研究（GWAS）分析指标对比。采用双侧Z检验比较HARE超级种群划分方案与1kGP+HGDP超级种群划分方案下的遗传力、LDSC截距（LDSC intercepts）及衰减比；采用5%的错误发现率进行多重检验校正。