Skip to content

[2022] Deep fusion learning facilitates anatomical therapeutic chemical recognition in drug repurposing and discovery #6

@inoue0426

Description

@inoue0426

Wang et al, 2021, Briefings in Bioinformatics.

  1. What kind of research(abstract)?

Drug repurposing and discovery using DeepATC.
ATC is the anatomical therapeutic chemical classification system, defined by WHO.
This method used graph convolutional network, inferring biological network and multimode attentive fusion network.

  1. What is progress from previous research?

  2. What is the specific topic?

  • dataset
    • Drugbank
    • KEGG
    • STITCH

Five drug networks were constructed according to iATC-NRAKEL.

Three types of chemical-chemical interaction were extracted using STITCH.

Two types of chemical-chemical similarity scores were obtained from KEGG using SIMCOMP and SUBCOMP.

Library of Integrated Network-Based Cellular Signatures (LINCS) is used for creating transcriptome data.

Benchmark dataset of Drug-Target Interaction was proposed according to A network integration approach for drug-target interaction prediction and computational drug repositioning from heterogeneous information.

The drug-drug interaction was extracted from DrugBank and the protein-protein interactions were obtained from the HPRD.

The drug-disease and protein-disease associations were extracted from Comparative Toxicogenomics Database .

The drug-side effect associations were extracted from the SIDER DB.

The structural similarity of drugs was calculated by the Tanimoto coefficient.

This model is constructed by three submodules

  • molecular semantic embedding (MSE) module
  • heterogeneous-network embedding (HNE) module
  • multimodel attentive fusion (MAF) module
    • ?

DeepATC: [WIP]

  1. How did the method evaluate its usefulness?

Compared with KNN, RF, and SVM.

  1. Are there any discussions?

  2. What is the next article you should read?

Metadata

Metadata

Assignees

No one assigned

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions