Design of overlapping genes using deep generative models of protein sequences

Area of Science:

• Molecular Biology
• Bioinformatics
• Synthetic Biology

Background:

• Viruses often utilize overlapping genes (OLG) in alternate reading frames, despite reduced protein sequence space.
• The degeneracy of amino acid sequences concerning protein folding and the feasibility of overlapping functional proteins remain open questions.

Purpose of the Study:

• To investigate whether amino acid sequences can accommodate functional protein folds in overlapping arrangements.
• To engineer synthetic OLGs using advanced generative models and assess their viability.

Main Methods:

• Designed overlapped sequences targeting two distinct protein families.
• Employed state-of-the-art generative models for sequence engineering.
• Encoded distinct, highly ordered de novo protein structures within synthetic OLGs.

Main Results:

• Achieved surprisingly high success rates in both in silico and experimental evaluations.
• Demonstrated that genetic code constraints do not significantly hinder simultaneous 3D fold accommodation in alternate reading frames.
• Confirmed the feasibility of engineering functional, overlapped protein structures.

Conclusions:

• Overlapping gene sequences are not significantly restricted by folding constraints under the standard genetic code.
• OLG sequences are likely frequently accessible in nature.
• Synthetic OLGs offer a promising strategy for compressing and constraining genetic circuits.