Skip to content

Welcome to DEM documentation!

pypi-badge pypi-badge License

Dual-Extraction Modeling (DEM): A multi-modal deep-learning architecture for phenotypic prediction and functional gene mining of complex traits

DOI: 10.1016/j.xplc.2024.101002

A multi-modal deep-learning architecture designed to extract representative features from diverse omics datasets. DEM enables robust and precise prediction of qualitative and quantitative traits phenotypes.

DEM architecture

Key Features:

  • Multi-Modal Deep Learning: DEM efficiently integrates heterogeneous omics data for both classification and regression tasks.
  • High Accuracy & Generalizability: Benchmarking experiments demonstrate DEM’s superior accuracy, robustness, and flexibility across various complex trait predictions.
  • Explainability: DEM excels at identifying pleiotropic genes, such as those influencing flowering time and rosette leaf number, with impressive Explainability.
  • User-Friendly Software: The repository includes easy-to-use tools for seamless application of DEM.

The DEM is implemented in the Python package biodem, which comprises 4 modules:

  • Data preprocessing
  • Dual-extraction modeling
  • Phenotypic prediction
  • Functional gene mining

modules of biodem