高精度肺癌亚型诊断不平衡的外体数据通过Exo-LCClassifier
Siyu Zhan1,2, Hao Yu3, Shuang Liu4
1Institute of Intelligent Computing, University of Electronic Science and Technology of China, Chengdu, Sichuan, China.
Frontiers in genetics
|May 15, 2025
概括
这项研究引入了Exo-LCClassifier,这是一种用于肺癌亚型预测的深度学习方法. 它有效地解决了不平衡的基因表达数据,实现了高诊断精度.
科学领域:
- 计算生物学 计算生物学
- 基因组学就是基因组学.
- 机器学习 机器学习
背景情况:
- 基因表达分析对于肺癌研究和亚型分化至关重要.
- 高维度和基因表达不平衡的数据带来了诊断挑战.
- 准确的肺癌亚型预测需要强大的分析方法.
研究的目的:
- 开发一种基于深度学习的创新方法,用于预测肺癌亚型.
- 为了克服高维度和不平衡的基因表达数据所带来的挑战.
- 使用分子特征提高肺癌诊断的准确性.
主要方法:
- 拟议的 Exo-LCClassifier 集成特征选择,1D CNN 和 WGAN.
- 使用DESeq2进行差异基因表达分析.
- 采用WGAN进行数据增强,以解决样本不平衡并增强样本多样性.
- 应用1D CNN用于平衡数据集的分类.
主要成果:
- 通过五倍交叉验证,通过五倍交叉验证实现了0.9766 ± 0.0070的平均准确性.
- 报告了高精度 (0.9762 ± 0.0101),回忆 (0.9827 ± 0.0050) 和F1分数 (0.9793 ± 0.0068) 的情况.
- 在外部GEO数据集上表现出强的表现,准确度为0.9588.
结论:
- 开发了一种创新的计算框架,以解决肺癌基因表达分析中的失衡学习问题.
- 集成了DESeq2,WGAN和1D CNN,用于有效的微分表达式分析,数据增强和分类.
- 源代码是公开的可供复制性和进一步研究.
相关概念视频
lncRNA - Long Non-coding RNAs
8.4K
In humans, more than 80% of the genome gets transcribed. However, only around 2% of the genome codes for proteins. The remaining part produces non-coding RNAs which includes ribosomal RNAs, transfer RNAs, telomerase RNAs, and regulatory RNAs, among other types. A large number of regulatory non-coding RNAs have been classified into two groups depending upon their length – small non-coding RNAs, such as microRNA, which are less than 200 nucleotides in length, and long non-coding RNA...
8.4K
Overview of Exosomes
2.7K
Exosomes are stable, lipid bilayer-enclosed vesicles capable of crossing biological barriers. They can carry a wide range of molecules required for intercellular communication. Once exosomes are released from the cell where they originated, they enter a recipient cell through various pathways such as fusion, receptor-mediated endocytosis, macropinocytosis, and phagocytosis.
Stahl et al. discovered exosomes in 1983, but the exosomes were initially considered waste products released from the...
Stahl et al. discovered exosomes in 1983, but the exosomes were initially considered waste products released from the...
2.7K


