Exploration of natural red-shifted rhodopsins using a machine learning-based Bayesian experimental design.

Area of Science:

• Biochemistry
• Molecular Biology
• Optogenetics

Background:

• Microbial rhodopsins are crucial photoreceptive proteins used as molecular tools in optogenetics.
• Developing new rhodopsins with specific spectral properties is essential for advancing optogenetic applications.

Purpose of the Study:

• To introduce a machine learning (ML)-based experimental design for screening microbial rhodopsins.
• To identify rhodopsins with red-shifted absorption spectra for enhanced optogenetic applications.

Main Methods:

• A machine learning model was developed to predict red-shifted rhodopsins within subfamilies.
• Protein BLAST searches identified 3,022 ion-pumping rhodopsins.
• The ML model selected 65 candidate rhodopsins for experimental validation.

Main Results:

• 39 candidate rhodopsins were expressed and their wavelengths determined.
• 82% (32 out of 39) showed significant red-shift gains.
• Four rhodopsins exhibited red-shift gains over 20 nm, with two possessing desirable ion-transporting properties.

Conclusions:

• Data-driven, ML-based approaches are effective for experimental design in rhodopsin research.
• The identified rhodopsins show potential for improved optogenetic tools.
• This strategy can accelerate discoveries in photobiology.