Amino Acid Grouping TSR

Overview

Amino acid grouping was introduced to improve the accuracy of identifying structurally conserved regions across proteins. By grouping amino acids according to their structural similarities, the method better captures global structural similarities and makes it easier to discover conserved motifs, which are important for understanding protein function.

Applied to the Triangular Spatial Relationship (TSR)-based method, amino acid grouping has been shown to modestly improve the accuracy of protein clustering in specific cases. It also aids in identifying key substructures, referred to as common and specific keys. Common keys represent substructures shared among different types of proteins, while specific keys belong to a particular protein type, giving deeper insight into structural relationships.

The work also demonstrates that amino acid grouping helps uncover conserved binding sites, such as those found in proteins involved in viral infections including SARS-CoV-2. These findings are relevant to antiviral drug design, where identifying critical structural features supports therapeutic targeting.

Run it with BioTSR

Amino acid grouping is an option of BioTSR's protein Cα method, --aa-grouping. To compute grouped keys for chain A of ubiquitin:

python -m pip install "biotsr==1.0.0"
biotsr fetch 1UBQ --pdb-dir structures
biotsr run --method protein --pdb structures/1UBQ.pdb --chain A --aa-grouping --output results

The run forms the same 70,300 triangles as an ungrouped one, but they produce 26,395 distinct keys rather than 45,349, because residues in the same group now share a label. The files are written under the prefix results/1UBQ_A_protein_aagroup.

• Which methods take it: the two protein Cα methods, protein and aa_selection (the Cα atoms of residues you list). Every other method refuses the option.
• From Python: biotsr.analyze(..., aa_grouping=True).
• Comparing results: grouped and ungrouped keys are different key spaces, and BioTSR will not compare a grouped result with an ungrouped one.

For many structures, a two-column CSV of protein IDs and chains, as the earlier package used, works as a BioTSR manifest; --download-missing fetches any structure not yet in the folder:

protein,chain
1GTA,A
1GTB,A
1LBE,A

biotsr batch --manifest sample_details.csv --method protein --aa-grouping --pdb-dir structures --download-missing --output results

Tutorial 01 runs grouping beside the other Cα options, and the user guide, §11 describes every option.

Grouped labels in BioTSR

BioTSR keeps the 13 groups of the standard amino acids. Selenomethionine (MSE) takes methionine's group, and each other supported modified residue keeps a label of its own: 23 grouped labels in all, encoded at radix 30. The earlier implementation encoded its 13 groups within the ungrouped 20-label space, so the same grouped triangle receives a different integer in the two programs; SSE-TSR §7 lists this with BioTSR's other conventions.