MOFNet: a deep learning framework for multi-omics data fusion in cancer subtype classification.

IF 2.4 4区生物学 Q3 BIOCHEMISTRY & MOLECULAR BIOLOGY

Molecular omics Pub Date : 2025-10-01 DOI:10.1039/d5mo00221d

Guangji Zhang, Chunxiao Zhang, Pengpai Li, Duanchen Sun, Zhixia Yang, Zhi-Ping Liu

{"title":"MOFNet: a deep learning framework for multi-omics data fusion in cancer subtype classification.","authors":"Guangji Zhang, Chunxiao Zhang, Pengpai Li, Duanchen Sun, Zhixia Yang, Zhi-Ping Liu","doi":"10.1039/d5mo00221d","DOIUrl":null,"url":null,"abstract":"Background: cancer exhibits high molecular and clinical heterogeneity, making accurate subtyping essential for personalized treatment. Traditional single-omics approaches often fail to capture this complexity. Multi-omics integration offers a more holistic understanding, but many existing methods either lack interpretability or fail to model cross-omics correlations effectively.Methods: we developed MOFNet, a novel supervised deep learning framework for multi-omics integration, incorporating a similarity graph pooling (SGO) module and a view correlation discovery network (VCDN). MOFNet processes omics data-including mRNA expression, DNA methylation, and miRNA expression-via omics-specific graph learning and cross-omics label space fusion. Three cancer types-breast cancer (BRCA), low-grade glioma (LGG), and stomach adenocarcinoma (STAD)-were analyzed using datasets from the cancer genome atlas (TCGA). Statistical evaluation was performed using accuracy, weighted F1 score, and macro F1 score across stratified training/testing splits.Results: MOFNet achieved superior performance across all datasets. For BRCA, it obtained an accuracy of 85.17%, F1_weighted of 85.36%, and macro F1 of 80.93%, outperforming all baseline models by up to 18.25%. In LGG and STAD, MOFNet also showed robust gains, with maximum improvements of 23.72% and 21.56%, respectively. Omics ablation studies demonstrated enhanced performance with multi-omics integration. Functional enrichment analysis revealed that MOFNet-identified key features were involved in biologically relevant pathways such as cell cycle regulation, synaptic signaling, and ion transport.Conclusions: MOFNet enables scalable and interpretable multi-omics data fusion for cancer subtype classification, significantly improving predictive accuracy while retaining only 25% of input features. The integration of SGO and VCDN modules offers both biological interpretability and computational efficiency. These results suggest MOFNet's promising application in precision oncology and biomarker discovery.","PeriodicalId":19065,"journal":{"name":"Molecular omics","volume":" ","pages":""},"PeriodicalIF":2.4000,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Molecular omics","FirstCategoryId":"99","ListUrlMain":"https://doi.org/10.1039/d5mo00221d","RegionNum":4,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"BIOCHEMISTRY & MOLECULAR BIOLOGY","Score":null,"Total":0}

引用次数: 0

Abstract

Background: cancer exhibits high molecular and clinical heterogeneity, making accurate subtyping essential for personalized treatment. Traditional single-omics approaches often fail to capture this complexity. Multi-omics integration offers a more holistic understanding, but many existing methods either lack interpretability or fail to model cross-omics correlations effectively.

Methods: we developed MOFNet, a novel supervised deep learning framework for multi-omics integration, incorporating a similarity graph pooling (SGO) module and a view correlation discovery network (VCDN). MOFNet processes omics data-including mRNA expression, DNA methylation, and miRNA expression-via omics-specific graph learning and cross-omics label space fusion. Three cancer types-breast cancer (BRCA), low-grade glioma (LGG), and stomach adenocarcinoma (STAD)-were analyzed using datasets from the cancer genome atlas (TCGA). Statistical evaluation was performed using accuracy, weighted F1 score, and macro F1 score across stratified training/testing splits.

Results: MOFNet achieved superior performance across all datasets. For BRCA, it obtained an accuracy of 85.17%, F1_weighted of 85.36%, and macro F1 of 80.93%, outperforming all baseline models by up to 18.25%. In LGG and STAD, MOFNet also showed robust gains, with maximum improvements of 23.72% and 21.56%, respectively. Omics ablation studies demonstrated enhanced performance with multi-omics integration. Functional enrichment analysis revealed that MOFNet-identified key features were involved in biologically relevant pathways such as cell cycle regulation, synaptic signaling, and ion transport.

Conclusions: MOFNet enables scalable and interpretable multi-omics data fusion for cancer subtype classification, significantly improving predictive accuracy while retaining only 25% of input features. The integration of SGO and VCDN modules offers both biological interpretability and computational efficiency. These results suggest MOFNet's promising application in precision oncology and biomarker discovery.

查看原文本刊更多论文

MOFNet：用于癌症亚型分类中多组学数据融合的深度学习框架。

背景：癌症表现出高度的分子和临床异质性，使得准确的亚型对个性化治疗至关重要。传统的单组学方法往往无法捕捉到这种复杂性。多组学集成提供了更全面的理解，但许多现有的方法要么缺乏可解释性，要么不能有效地模拟跨组学的相关性。方法：我们开发了MOFNet，这是一个新的多组学集成监督深度学习框架，结合了相似图池（SGO）模块和视图关联发现网络（VCDN）。MOFNet通过组学特异性图学习和跨组学标签空间融合处理组学数据，包括mRNA表达、DNA甲基化和miRNA表达。使用癌症基因组图谱（TCGA）的数据集分析了三种癌症类型——乳腺癌（BRCA）、低级别胶质瘤（LGG）和胃腺癌（STAD）。采用准确性、加权F1分数和宏观F1分数对分层训练/测试分割进行统计评估。结果：MOFNet在所有数据集上都取得了卓越的性能。对于BRCA，其准确率为85.17%，F1_weighted为85.36%，macro F1为80.93%，优于所有基线模型高达18.25%。在LGG和STAD中，MOFNet也表现出强劲的增长，最大增幅分别为23.72%和21.56%。组学消融研究表明，多组学整合可以提高性能。功能富集分析显示，mofnet鉴定的关键特征涉及生物学相关途径，如细胞周期调节、突触信号传导和离子运输。结论：MOFNet为癌症亚型分类提供了可扩展和可解释的多组学数据融合，在仅保留25%输入特征的情况下显著提高了预测准确性。SGO和VCDN模块的集成提供了生物可解释性和计算效率。这些结果表明MOFNet在精确肿瘤学和生物标志物发现方面具有广阔的应用前景。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Molecular omics Biochemistry, Genetics and Molecular Biology-Biochemistry

CiteScore

5.40

自引率

3.40%

发文量

期刊介绍： Molecular Omics publishes high-quality research from across the -omics sciences. Topics include, but are not limited to: -omics studies to gain mechanistic insight into biological processes – for example, determining the mode of action of a drug or the basis of a particular phenotype, such as drought tolerance -omics studies for clinical applications with validation, such as finding biomarkers for diagnostics or potential new drug targets -omics studies looking at the sub-cellular make-up of cells – for example, the subcellular localisation of certain proteins or post-translational modifications or new imaging techniques -studies presenting new methods and tools to support omics studies, including new spectroscopic/chromatographic techniques, chip-based/array technologies and new classification/data analysis techniques. New methods should be proven and demonstrate an advance in the field. Molecular Omics only accepts articles of high importance and interest that provide significant new insight into important chemical or biological problems. This could be fundamental research that significantly increases understanding or research that demonstrates clear functional benefits. Papers reporting new results that could be routinely predicted, do not show a significant improvement over known research, or are of interest only to the specialist in the area are not suitable for publication in Molecular Omics.