Early detection of hepatocellular carcinoma via no end-repair enzymatic methylation sequencing of cell-free DNA and pre-trained neural network, Genome Med, 2023. (IF: 10.4)

Abstract: In this study, a low DNA damage and high-fidelity methylation detection method called No End-repair Enzymatic Methyl-seq (NEEM-seq) was developed, along with a detection model named DeepTrace, to better identify sequencing reads derived from hepatocellular carcinoma (HCC) through a pre-trained and fine-tuned neural network. The study demonstrated that NEEM-seq overcomes the shortcomings of traditional enzymatic methylation sequencing methods by preventing the introduction of unmethylation errors in cell-free DNA (cfDNA). DeepTrace outperformed other models in identifying HCC-derived reads and detecting HCC individuals. The research indicates that the combination of high-fidelity methylation data from NEEM-seq with the DeepTrace model holds great potential for early HCC detection, with high sensitivity and specificity, making it potentially suitable for clinical applications.

摘要：在这项研究中，开发了一种名为No End-repair Enzymatic Methyl-seq (NEEM-seq)的低DNA损伤和高保真度的甲基化检测方法，以及DeepTrace的检测模型，用于通过预训练和微调的神经网络更好地识别源自肝细胞癌（HCC）的测序读段。研究结果显示，NEEM-seq通过避免在cfDNA中引入未甲基化错误，克服了传统酶促甲基化测序方法的缺点。DeepTrace在识别HCC衍生读段和检测HCC个体方面的表现优于其他模型。研究指出，结合NEEM-seq的高保真度甲基化数据和DeepTrace模型，该方法在HCC早期检测方面具有很高的潜力，具有高灵敏度和特异性，可能适合临床应用。