MOKPE: drug-target interaction prediction via manifold optimization based kernel preserving embedding.

Area of Science:

• Bioinformatics
• Computational Biology
• Drug Discovery

Background:

• Bioinformatics often involves integrating data from diverse, heterogeneous sources.
• Identifying drug-target interactions (DTIs) is crucial for drug discovery.
• Existing methods struggle with modeling complex, heterogeneous data for DTI prediction.

Purpose of the Study:

• To propose a novel framework, Manifold Optimization based Kernel Preserving Embedding (MOKPE), for modeling heterogeneous data.
• To efficiently predict drug-target interactions by preserving similarities within and between data types.
• To enhance the accuracy and efficiency of DTI identification in drug discovery.

Main Methods:

• Developed the Manifold Optimization based Kernel Preserving Embedding (MOKPE) framework.
• Projected heterogeneous drug and target data into a unified embedding space.
• Preserved drug-target interactions, drug-drug similarities, and target-target similarities simultaneously.

Main Results:

• MOKPE demonstrated superior or comparable performance in predicting DTIs compared to state-of-the-art similarity-based methods.
• Evaluated through ten replications of ten-fold cross-validation on four DTI network datasets.
• Successfully predicted previously unseen DTIs and evaluated on predicting unknown DTIs within a given network.

Conclusions:

• MOKPE is an effective framework for modeling heterogeneous data in bioinformatics.
• The proposed method significantly advances DTI prediction accuracy and efficiency.
• The R implementation and replication scripts are publicly available for further research.