ExactCN: Predicting Exact Copy Numbers on Whole Exome Sequencing Data

The quantification of the precise copy number variations (CNVs) is crucial to understandingthe effects of gene dosage, disease severity, and therapeutic response. Although whole-exome sequencing(WES) offers a cost-effective solution for CNV detection in a clinical setting, it introduces several biases,including those related to sequence length, GC content, and the use of targeting probes. Consequently,estimating exact copy numbers remains challenging, especially for WES data. Here, we present Ex-actCN, a deep learning–based method for estimation of exact copy numbers from WES data per exon.The architecture integrates convolutional layers that extract local read-depth patterns with transformerencoder blocks that capture genomic context and handle sequencing noise. ExactCN is trained on WESsamples from the 1000 Genomes Project, using matching WGS-based calls as semi–ground truth. Inbenchmarks, ExactCN improves the state-of-the-art integer CNV calling performance by reducing themacro-averaged mean absolute error (MAE) from 0.91 to 0.62 and the macro-averaged root meansquared error (RMSE) from 1.31 to 0.78. It also achieves an overall Pearson correlation of 0.669 andSpearman correlation of 0.550, improving the second-best method by 0.641 and 0.482, respectively.Furthermore, a fine-tuned version of ExactCN demonstrated an overall F1-score of 0.657 for ag-gregate CNV detection performance on the clinically important duplicated genes SMN1/2, demon-strating its applicability to both research and clinical genomic analyses.精准定量拷贝数变异（copy number variations, CNVs）对于理解基因剂量效应、疾病严重程度以及治疗反应至关重要。尽管全外显子测序（whole-exome sequencing, WES）为临床场景下的CNV检测提供了一种经济高效的解决方案，但该技术会引入多种偏倚，包括与序列长度、GC含量以及靶向探针使用相关的偏倚。因此，精准估算拷贝数仍颇具挑战，针对WES数据的此类任务更是如此。为此，我们提出ExactCN——一种基于深度学习的方法，可从WES数据中针对每个外显子估算精准拷贝数。该模型架构整合了用于提取局部测序读段深度模式的卷积层，以及用于捕获基因组上下文并处理测序噪声的Transformer编码器模块。ExactCN以1000基因组计划（1000 Genomes Project）的WES样本作为训练数据，并以匹配的全基因组测序（whole genome sequencing, WGS）得到的拷贝数调用结果作为半真值标签。在基准测试中，ExactCN优化了当前最先进的整数型CNV调用性能：将宏平均绝对误差（macro-averaged mean absolute error, MAE）从0.91降至0.62，将宏平均均方根误差（macro-averaged root mean squared error, RMSE）从1.31降至0.78。此外，该模型整体皮尔逊相关系数达0.669，斯皮尔曼相关系数达0.550，分别较次优方法提升了0.641与0.482。进一步研究表明，经过微调的ExactCN版本在临床重要的运动神经元生存基因SMN1/2的聚集性CNV检测任务中，整体F1分数可达0.657，证明其可应用于科研与临床基因组分析。