大型语言模型在利用细胞游离DNA末端序列特征进行癌症高精度诊断中的应用

Large language model produces high accurate diagnosis of cancer from end-motif profiles of cell-free DNA

DOI 原文链接

用sci-hub下载

BRIEFINGS IN BIOINFORMATICS

影响因子:7.7

分区:生物学2区 / 数学与计算生物学1区生化研究方法2区

发表日期:2024 Jul 25

作者: Jilei Liu, Hongru Shen, Kexin Chen, Xiangchun Li

DOI: 10.1093/bib/bbae430

摘要

经过指令调优的大型语言模型（LLMs）展现出极高的与人类意图对齐能力。我们提出一种基于LLM的模型——iLLMAC（基于指令调优的癌症检测模型），可以利用细胞游离DNA（cfDNA）末端序列特征检测癌症。在来自1135名癌症患者和1106名对照者的血浆cfDNA测序数据基础上开发，iLLMAC在三组数据中实现了癌症诊断的受试者工作特征面积（AUROC）为0.866（95%置信区间CI，0.773-0.959），利用16个末端序列特征检测肝细胞癌（HCC）的AUROC为0.924（CI，0.841-1.0）。随着特征数增加，性能提升至癌症诊断的0.886（CI，0.794-0.977）和HCC检测的0.956（CI，0.89-1.0），使用64个末端序列特征。在外部测试集中，64个末端序列特征的癌症诊断AUROC达到0.912（CI，0.849-0.976），HCC检测AUROC为0.938（CI，0.885-0.992），显著优于其他基准方法。此外，iLLMAC在甲硫氨酸全基因组甲基化和5-羟甲基胞嘧啶测序数据集上也表现出高分类性能。我们的研究强调了基于LLM的指令调优在cfDNA癌症检测中的有效性。

Abstract

Instruction-tuned large language models (LLMs) demonstrate exceptional ability to align with human intentions. We present an LLM-based model-instruction-tuned LLM for assessment of cancer (iLLMAC)-that can detect cancer using cell-free deoxyribonucleic acid (cfDNA) end-motif profiles. Developed on plasma cfDNA sequencing data from 1135 cancer patients and 1106 controls across three datasets, iLLMAC achieved area under the receiver operating curve (AUROC) of 0.866 [95% confidence interval (CI), 0.773-0.959] for cancer diagnosis and 0.924 (95% CI, 0.841-1.0) for hepatocellular carcinoma (HCC) detection using 16 end-motifs. Performance increased with more motifs, reaching 0.886 (95% CI, 0.794-0.977) and 0.956 (95% CI, 0.89-1.0) for cancer diagnosis and HCC detection, respectively, with 64 end-motifs. On an external-testing set, iLLMAC achieved AUROC of 0.912 (95% CI, 0.849-0.976) for cancer diagnosis and 0.938 (95% CI, 0.885-0.992) for HCC detection with 64 end-motifs, significantly outperforming benchmarked methods. Furthermore, iLLMAC achieved high classification performance on datasets with bisulfite and 5-hydroxymethylcytosine sequencing. Our study highlights the effectiveness of LLM-based instruction-tuning for cfDNA-based cancer detection.