文献速递肝病

一种熵权法整合大组学数据并机制性评价药物性肝损伤

An entropy weight method to integrate big omics and mechanistically evaluate DILI.
PubMed 文献 · Hepatology · 2023-10-11

本研究整合大鼠肝脏基因表达谱、FDA 肝毒性知识库药物标签和 LiverTox 临床报告等多源数据,基于毒性通路建立 11 条通路组合,采用熵权法加权通路内响应基因并训练可解释机器学习模型 XGBoot-SHAP 排序通路重要性;在 89 种药物(61 阳性+28 阴性)测试中精确度可达 86% 及以上。

要点

  • 基于 11 条通路组合的模型可区分损伤与非损伤样本点并刻画 DILI 表现。
  • 熵权法加权通路内响应基因,XGBoot-SHAP 模型排序通路对组合活性的重要性。
  • 测试 89 种药物(61 阳性+28 阴性)精确度可达 86% 及以上。

核心结论

  • 本研究提供了一种机制驱动的预测建模新方法及大数据整合思路,有助于药理学和其他人类生物学领域。

研究设计与样本

  • 整合开放 TG-GATE 数据库和 DrugMatrix 的大鼠肝脏基因表达谱、FDA 肝毒性知识库药物标签及 LiverTox 临床报告等多源数据集,运用生物信息学和计算工具开展研究。
  • 使用开放 TG-GATEs 中 6 种阳性与 1 种阴性 DILI 药物作为训练集建立 11 条通路组合。
  • 采用熵权法加权通路内响应基因,训练可解释机器学习模型 XGBoot-SHAP 排序通路对组合活性的重要性。
  • 使用额外 89 种药物(61 阳性+28 阴性)测试模型。

主要发现

  • 建立的 11 条通路组合广泛覆盖生物过程和应激反应。
  • 组合活性可区分损伤与非损伤样本点,并利用 6 种训练药物刻画 DILI 表现。
  • 在 89 种药物(61 阳性+28 阴性)测试中,精确度可达 86% 及以上。

局限性

  • 原文未报告局限性

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

BACKGROUND AND AIMS: DILI accounts for more than half of acute liver failure cases in the United States and is a major health care issue for the public worldwide. As investigative toxicology is playing an evolving role in the pharmaceutical industry, mechanistic insights into drug hepatotoxicity can facilitate drug development and clinical medication. METHODS: By integrating multisource datasets including gene expression profiles of rat livers from open TG-GATE database and DrugMatrix, drug labels from FDA Liver Toxicity Knowledge Base, and clinical reports from LiverTox, and with the employment of bioinformatic and computational tools, this study developed an approach to characterize and predict DILI based on the molecular understanding of the processes (toxicity pathways). RESULTS: A panel of 11 pathways widely covering biological processes and stress responses was established using a training set of six positive and one negative DILI drugs from open TG-GATEs. An entropy weight method-based model was developed to weight responsive genes within a pathway, and an interpretable machine-learning (ML) model XGBoot-SHAP was trained to rank the importance of pathways to the panel activity. The panel activity was proven to differentiate between injured and noninjured sample points and characterize DILI manifestation using six training drugs. Next, the model was tested using an additional 89 drugs (61 positives + 28 negatives), and a precision of 86% and higher can be achieved. CONCLUSIONS: This study provides a novel approach to mechanisms-driven prediction modeling, as well as big data integration for insights into pharmacology and other human biology areas.

来源:Hepatology,查看原文。版权归原作者所有。

引用

  1. Jin Y, Shou Y, Lei Q, et al. An entropy weight method to integrate big omics and mechanistically evaluate DILI. Hepatology. 2023. PMID: 37820269. doi:10.1097/HEP.0000000000000628

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。