Guiding discovery of protein sequence-structure-function modeling.

Area of Science:

• Biochemistry and Molecular Biology
• Computational Biology
• Protein Engineering

Background:

• Protein engineering is crucial for developing novel catalysts.
• Current methods are limited by laborious protein expression and screening.
• Understanding sequence-structure-function relationships is key but challenging.

Purpose of the Study:

• To develop and test a high-throughput in silico pipeline for predicting protein function.
• To accelerate the design and engineering of novel protein catalysts.
• To benchmark the pipeline using fungal flavin-dependent monooxygenases.

Main Methods:

• Utilized AlphaFold2 for structure prediction and fast Fourier transform docking.
• Developed a sequence-structure-function pipeline for in silico screening.
• Employed ensemble decision tree models and explainable AI for sequence-function modeling.

Main Results:

• Pipeline predictions for enantioselectivity and reactivity correlated well with experimental data.
• Identified key residues controlling protein enantioselectivity and reactivity using AI.
• Validated identified residues against experimentally verified sites.

Conclusions:

• The in silico pipeline effectively accelerates protein engineering efforts.
• This approach enables exploration of vast sequence landscapes for catalyst design.
• The developed pipeline provides a framework for informed protein engineering from sequence data.