Features Collected from Amino Acids for Novel PHA Synthase Design

Approximately 10,000 novel PHA synthase enzyme sequences were designed using a conditional variational autoencoder trained on native protein sequences from Uniprot. Tuula Tenkanen authored this dataset, which was last updated on March 19, 2026. The dataset likely contains features collected from amino acids, used to select 16 sequences for in vivo validation.

Use Cases

• Training machine learning models for protein sequence generation based on conserved residues and structural features.
• Selecting novel enzyme candidates for experimental validation based on criteria like catalytic amino acids and dimer interface residues.
• Analyzing the relationship between amino acid substitutions and enzyme activity, as two novel enzymes with 87 and 98 substitutions were confirmed active.
• Studying structural features like the number of α-helices in the N-terminal part of enzymes for functional prediction.

Strengths

• Dataset includes features derived from approximately 10,000 novel enzyme sequences designed via AI.
• 16 sequences were selected based on specific, documented criteria including conserved residues and structural features.
• Two novel enzymes were experimentally confirmed to be active, providing a validation benchmark.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset is 5.5 KB, indicating a very limited scope of data.

Provenance

Source

Native protein sequences obtained from Uniprot.

Collection Method

Designed using a conditional variational autoencoder.

Freshness

Last updated 2026-03-19 17:34:40; freshness should be verified.