A review of Deep Learning methods for multi-omics data integration
Year: 2026
Type: article
Abstract: The rapid advancement of high-throughput sequencing technologies has led to an exponential growth in omics data， offering unprecedented opportunities for analysis of complex biological mechanisms. The integration of multi-omics data， which combines diverse biological layers such as genomics， transcriptomics， proteomics， and metabolomics， has emerged as a pivotal approach for characterizing the molecular mechanisms of diseases and advancing biomedical research. However， the integration of these heterogeneous data sources presents significant challenges stemming from issues including high dimensionality， missing values， and inter-modality differences. Traditional integration methods have limitations in modeling non-linear interactions， often failing to fully exploit the latent information within omics data. In contrast， deep learning approaches have demonstrated remarkable potential to overcome these challenges， exhibiting superior performance in feature extraction， modality alignment， and data reconstruction. This paper provides a comprehensive review of the mainstream deep learning methods for multi-omics data integration， systematically evaluating their respective strengths， limitations， and applications in biomedical research.
Authors: LAN Wei, LI Wenhao, WEI Xin
Language: Chinese
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7162729259