Additional file 5: of Exome sequencing of primary breast cancers with paired metastatic lesions reveals metastasis-enriched mutations in the A-kinase anchoring protein family (AKAPs)

Figures S3a and S3b. Estimation of LOH fraction and tumor content. Estimation of LOH fraction and tumor content using a Beta-Normal mixture model. Histograms show the distribution of major allele frequencies (range: 50–100%) and red curves show the mixture model estimated from the data. The set of SNPs called in the germline sample was considered in the tumor sample (primary and metastasis independently). When genomes contain regions of LOH, the distribution will be bimodal, as illustrated in the inset (top right). The first component (closer to 50%; red in inset) represents heterozygous SNPs in regions without LOH and was modeled as a normal distribution with a mean close to 50% in a perfect sample, but will increase towards 100% as allelic dropout increases. The second component (closer to 100%; blue in inset) represents homozygous SNPs in regions with LOH, was modeled as a beta distribution with two parameters (the mean of this component should be close to 100% and increase as LOH fraction increases and decrease as with non-cancer cell contamination. The mixture distribution thus had four free parameters plus the mixture proportion, the latter representing the estimated LOH fraction of the sample. Fitting this mixture to the observations yielded estimates for LOH fraction, tumor fraction and allelic dropout rate for each sample (Additional file 3: Figure S2). Model fitting was performed using the EstimatedDistribution function of Mathematica 9.0 (Wolfram Research Inc.). (ZIP 421 kb)

补充图S3a与S3b：杂合性缺失（Loss of Heterozygosity, LOH）比例与肿瘤细胞含量估算。本部分介绍基于Beta-正态混合模型的杂合性缺失比例与肿瘤细胞含量估算方法。直方图呈现主要等位基因频率（范围：50%~100%）的分布特征，红色曲线为基于实验数据拟合得到的混合模型结果。我们在肿瘤样本（分别独立针对原发灶与转移灶）中纳入了在种系样本中检测得到的单核苷酸多态性（Single Nucleotide Polymorphisms, SNPs）位点集。当基因组存在杂合性缺失区域时，该等位基因频率分布将呈现双峰模式，如右上角内嵌图所示。第一个组分（峰值接近50%，内嵌图中为红色）代表无杂合性缺失区域内的杂合性SNPs，在理想实验样本中其分布可拟合为均值接近50%的正态分布，但随着等位基因脱扣现象加剧，该分布均值将向100%偏移。第二个组分（峰值接近100%，内嵌图中为蓝色）代表存在杂合性缺失区域内的纯合性SNPs，拟合为双参数Beta分布，该组分的均值应接近100%，且随杂合性缺失比例升高而增大，随非癌细胞污染程度升高而降低。因此该混合分布共包含4个自由参数与1个混合比例参数，其中混合比例参数即为本样本的杂合性缺失比例估算值。将该混合模型与观测数据拟合后，可得到每个样本的杂合性缺失比例、肿瘤细胞占比与等位基因脱扣率估算值（详见补充文件3的补充图S2）。模型拟合通过Mathematica 9.0（Wolfram Research公司）的EstimatedDistribution函数完成。（压缩包大小：421 KB）