Predicting Drug-Target Interactions with Electrotopological State Fingerprints and Amphiphilic Pseudo Amino Acid...

Area of Science:

• Computational Chemistry
• Bioinformatics
• Drug Discovery

Background:

• Drug-target interaction (DTI) prediction is crucial for efficient drug development.
• Experimental DTI identification is costly, time-consuming, and challenging.
• Machine learning methods for DTI prediction are limited by feature extraction and negative sampling strategies.

Purpose of the Study:

• To develop an effective and efficient machine learning model for DTI prediction.
• To explore the utility of electrotopological state (E-state) fingerprints for drugs and amphiphilic pseudo amino acid composition (APAAC) for target proteins.
• To propose a novel distance-based negative sampling method to improve prediction accuracy.

Main Methods:

• Utilized E-state fingerprints capturing molecular electronic and topological features for drugs.
• Employed APAAC, an extension of amino acid composition, incorporating hydrophilic/hydrophobic characters for target proteins.
• Developed a prediction model using support vector machines (SVM) with combined E-state and APAAC features.
• Implemented a distance-based negative sampling strategy to generate reliable negative samples.

Main Results:

• Achieved Area Under the Curve (AUC) values exceeding 98.5% across three independent datasets.
• Demonstrated superior performance compared to existing state-of-the-art DTI prediction methods.
• Validated the effectiveness and efficiency of the proposed feature combination and negative sampling approach.

Conclusions:

• The proposed machine learning framework integrating E-state fingerprints and APAAC with distance-based negative sampling significantly enhances DTI prediction accuracy.
• This method offers a computationally efficient and effective alternative to experimental approaches, facilitating accelerated drug discovery and development.
• The findings provide a valuable tool for researchers aiming to identify novel drug-target interactions.