Discovering the perfect drug
Problem being addressed
The drug discovery stage is a vital part of the drug development process and forms part of the initial stages of the development pipeline. In recent times, machine learning-based methods are actively being used to model drug-target interactions for rational drug discovery; it has led to having several descriptors for both targets and compounds, which adds challenge to both performance and interpretability of models.
A multi-view attention-based architecture for learning the representation of compounds and targets from different unimodal descriptor schemes (including end-to-end schemes) for drug-target interaction prediction. This usage of neural attention enables the proposed approach to lend itself to the interpretation and discovery of biologically plausible insights in compound-target interactions across multiple views.
Advantages of this solution
Experimental results demonstrate the ability of the suggested method to achieve high accuracy and offer biologically plausible interpretations using neural attention.
Solution originally applied in these industries
Possible New Application of the Work
The use of computer-based methods to optimize the drug development process can reduce healthcare costs and encourage accessibility of healthcare services.
Source URL: #############