PHA Synthase Enzyme Sequences Designed by AI for Validation

A conditional variational autoencoder was used to design approximately 10,000 novel polyhydroxyalkanoate synthase enzymes. 16 sequences were selected for in vivo validation based on conserved residues and structural features. Two novel enzymes with 87 and 98 amino acid substitutions were confirmed active in yeast.

Use Cases

• Benchmarking AI-generated protein sequences against native enzymes based on conserved residues.
• Studying the relationship between enzyme structure and function based on features like helix count.
• Training models for enzyme activity prediction based on sequence and substitution data.
• Investigating the impact of amino acid substitutions on enzyme dimerization and catalysis.

Strengths

• Includes 16 sequences selected for validation with specific selection criteria.
• Two novel enzymes were confirmed active, providing concrete validation results.
• Dataset is associated with a published study using a conditional variational autoencoder.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

figshare

Collection Method

Sequences were generated by a conditional variational autoencoder trained on native protein sequences from Uniprot.