Nonlinear data fusion over Entity-Relation graphs for Drug-Target Interaction prediction.

Area of Science:

• Computational chemistry
• Bioinformatics
• Machine learning

Background:

• Predicting Drug-Target Interactions (DTIs) is crucial for drug design and repurposing.
• Existing methods often lack flexibility or require stricter validation.

Purpose of the Study:

• To present a novel data fusion approach for DTI prediction using the NXTfusion library.
• To generalize Matrix Factorization for nonlinear inference over Entity-Relation graphs.

Main Methods:

• Developed a data fusion approach for DTI prediction.
• Extended Matrix Factorization to nonlinear inference on Entity-Relation graphs using NXTfusion.
• Benchmarked against state-of-the-art methods on five datasets.

Main Results:

• The proposed models outperformed existing methods on DTI prediction tasks.
• The approach demonstrated flexibility in predicting both binary DTIs and real-valued drug-target affinity.
• Findings suggest stricter validation for DTI methods in realistic settings.

Conclusions:

• The Entity-Relation data fusion approach effectively integrates heterogeneous information for DTI prediction.
• The method offers a flexible and high-performing solution for computer-aided drug design.
• Stricter validation mimicking real-world scenarios is recommended for DTI prediction methods.