Sonification based de novo protein design using artificial intelligence, structure prediction, and analysis using...

Area of Science:

• Computational Biology
• Protein Engineering
• Artificial Intelligence

Background:

• Designing novel proteins with specific functions is a significant challenge in biotechnology.
• Current methods often lack the ability to create truly unique and functional protein structures.

Purpose of the Study:

• To develop and validate a deep learning model for the de novo design of proteins.
• To translate protein sequence and structure information into a musical representation for AI-driven design.

Main Methods:

• A deep neural network utilizing long short-term memory units was trained on musical representations of proteins.
• The model translates protein data into musical scores, then generates new scores to design amino acid sequences.
• Protein structure prediction and validation were performed using established computational tools (ORION, MODELLER) and molecular dynamics.

Main Results:

• The deep learning model successfully designed novel de novo proteins not previously found in nature.
• Designed proteins demonstrated folding into specified secondary structures, confirmed by computational analysis.
• Molecular dynamics simulations validated the stability and characteristics of the newly designed proteins.

Conclusions:

• This innovative method enables the creation of unique protein sequences and structures.
• The approach offers a powerful new tool for designing protein-based materials for biology, medicine, and engineering.