CEMDA

Combined Embedding Model for Predicting MiRNA-Disease AssociationsCEMDA () is a tool for predicting potential associations between miRNAs (small non-coding RNA molecules) and diseases. It aids in understanding disease mechanisms at a molecular level and could contribute to improved disease diagnosis and treatment.

Methodology: Heterogeneous Network: CEMDA builds a network using known miRNA-disease associations, disease similarity data, and miRNA functional similarity.

Meta-Paths and GRU: It analyzes context paths within the network using Gate Recurrent Units (GRU) to learn similarity measures between miRNAs and diseases.

Attention Mechanism: Multi-head attention weighs the importance of different meta-paths, improving the accuracy of similar information.

Pair Embedding: Employs MLP to focus on key aspects of the pairwise relationship between miRNAs and diseases.

Combined Predictions: Combines meta-path-based and pair embedding predictions for more reliable miRNA-disease association predictions.

Topic

Functional, regulatory and non-coding RNA;Infectious disease;Microarray experiment;Oncology;Cardiology

Detail

  • Operation: miRNA target prediction;Network analysis;Aggregation

  • Software interface: Command-line interface

  • Language: Python

  • License: Not stated

  • Cost: Free of charge

  • Version name: -

  • Credit: China University of Mining and Technology.

  • Input: -

  • Output: -

  • Contact: Lei Zhang zhanglei@cumt.edu.cn ,Zhengwei Li zwli@cumt.edu.cn

  • Collection: -

  • Maturity: -

Publications

  • Combined embedding model for MiRNA-disease association prediction.
  • Liu B, et al. Combined embedding model for MiRNA-disease association prediction. Combined embedding model for MiRNA-disease association prediction. 2021; 22:161. doi: 10.1186/s12859-021-04092-w
  • https://doi.org/10.1186/S12859-021-04092-W
  • PMID: 33765909
  • PMC: PMC7995599

Download and documentation

< Back to DB search