A deep learning framework for integrating single-cell multi-omics data using graph-linked unified embeddings.
GLUE (Graph-Linked Unified Embedding) is a deep learning framework for integrating single-cell multi-omics data. It solves the problem of analyzing heterogeneous biological data from the same cells by learning a unified representation across different modalities like transcriptomics and epigenomics. The framework uses graph neural networks to model relationships between cells and features while preserving the specific characteristics of each data type.
Bioinformaticians and computational biologists working with single-cell multi-omics data who need to integrate different data modalities for comprehensive analysis. Researchers studying cellular heterogeneity and regulatory mechanisms across multiple molecular layers.
Developers choose GLUE because it provides a principled graph-based approach to multi-omics integration that preserves biological relationships, offers scalable performance for large datasets, and comes with reproducible workflows and comprehensive documentation. Unlike simpler integration methods, it models the complex relationships between different omics layers through graph neural networks.
Graph-linked unified embedding for single-cell multi-omics data integration
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Uses graph neural networks to model biological relationships between cells and features, preserving intrinsic data structures as shown in the model architecture diagram.
Supports integration of diverse single-cell data types like transcriptomics and epigenomics, enabling comprehensive analysis across modalities.
Designed for large-scale datasets with GPU acceleration options, ensuring efficient processing as mentioned in the installation commands.
Provides conda environment specifications and detailed reproduction instructions, including R environment setup via packrat, for reliable research.
Requires managing both Python and R environments with specific versions, as outlined in the reproduction steps, which can be cumbersome and error-prone for deployment.
Training graph neural networks on large datasets necessitates significant GPU resources, limiting accessibility for teams without high-performance computing infrastructure.
Assumes advanced knowledge in deep learning and bioinformatics, with fewer beginner-friendly resources compared to simpler integration tools like Seurat.