DeepStackVEGF a stacking ensemble deep learning framework for vascular endothelial growth factor prediction.

Area of Science:

• Biotechnology
• Computational Biology
• Bioinformatics

Background:

• Vascular Endothelial Growth Factor (VEGF) is crucial for angiogenesis, impacting wound healing, tissue repair, bone formation, and diseases like cancer and diabetic retinopathy.
• VEGF is a key therapeutic target for anti-angiogenic drugs and precision medicine.
• Current experimental methods for VEGF identification are expensive and time-consuming, necessitating efficient computational solutions.

Purpose of the Study:

• To develop DeepStack-VEGF, a deep learning framework for accurate and robust prediction of VEGF.
• To integrate diverse sequence-derived features and pretrained embeddings for enhanced prediction capabilities.
• To validate the efficacy of the proposed computational framework in supporting drug discovery and precision medicine.

Main Methods:

• Integration of sequence-derived features: physicochemical descriptors, sequential patterns, evolutionary information, and secondary structure motifs.
• Utilization of pretrained embeddings from UniProt and ProtBert.
• Feature optimization using Support Vector Machine-Recursive Feature Elimination.
• Employing a stacking ensemble of Feedback Generative Adversarial Network, Gated Recurrent Unit, and Capsule Convolutional Neural Network architectures.

Main Results:

• The fused feature set and stacking ensemble significantly outperformed individual models.
• Achieved superior accuracy, robustness, and generalization in VEGF prediction.
• Demonstrated the effectiveness of combining deep learning with biological insights.

Conclusions:

• DeepStack-VEGF offers a reliable and scalable computational framework for VEGF identification.
• The developed model supports rational drug discovery and the design of anti-angiogenic therapies.
• This approach advances precision medicine applications by enabling efficient VEGF analysis.