Sounds interesting: can sonification help us design new proteins?

Area of Science:

• Biophysics
• Computational Biology
• Bioinformatics

Background:

• Sonification, the practice of converting scientific data into music, is an emerging field.
• Previous protein sonification efforts mapped amino acids to notes or used overall protein structure for composition.
• Analogies exist between hierarchical protein structures and musical forms.

Purpose of the Study:

• To develop a comprehensive strategy for mapping proteins to music, incorporating geometry, sequence, and secondary structure.
• To enable the inverse mapping of music back to protein data, including sequence, secondary structure, and folding.
• To explore the potential of machine learning for generating novel music translatable to new proteins.

Main Methods:

• Developed a comprehensive mapping strategy encoding protein geometry, amino acid sequence, and secondary structure.
• Implemented an inverse mapping process to retrieve protein information from musical data.
• Proposed training machine learning models on 'protein music' for de novo protein design.

Main Results:

• Created a more complete musical representation closely linked to the original protein structure.
• Successfully retrieved amino acid sequence, secondary structure, and folding from musical data.
• Demonstrated the potential for generating novel music translatable to new proteins with specific design features.

Conclusions:

• The comprehensive mapping strategy enhances the link between protein data and its musical representation.
• The inverse mapping facilitates the retrieval of detailed protein information from music.
• Machine learning applied to 'protein music' offers a promising avenue for advanced protein design and discovery.