Recent Applications of Deep Learning Methods on Evolution- and Contact-Based Protein Structure Prediction.

Area of Science:

• Computational biology
• Structural bioinformatics
• Machine learning

Background:

• Deep learning methods are revolutionizing scientific research, particularly in understanding protein systems.
• Protein structure prediction relies heavily on machine learning to interpret sequence and homology data for inter-residue contacts and organization.

Purpose of the Study:

• To explore recent applications of deep learning in protein structure prediction.
• To identify future opportunities for deep learning in discovering novel protein structures and functions.
• To assess the role of deep learning in guiding drug-target interactions.

Main Methods:

• Review of recent deep learning applications in protein structure prediction.
• Analysis of deep neural network performance in competitions like CASP13 and CASP14.
• Exploration of potential deep learning strategies for structure and function identification.

Main Results:

• Deep learning, especially deep neural networks, has shown significant impact in protein structure prediction, evidenced by performance in CASP competitions.
• Deep learning methods are increasingly vital for interpreting complex protein sequence and homology data.
• Emerging opportunities exist for deep learning in uncovering unknown protein structures and functions.

Conclusions:

• Deep learning is a transformative technology in protein structural bioinformatics.
• Future applications of deep learning are expected to accelerate drug discovery and development.
• Despite challenges, deep learning will play a pivotal role in advancing protein science.