Amino Acid TSR

Overview

Amino Acid TSR generates TSR keys and triplets at the level of individual amino acids rather than whole protein chains: the points of its triangles are the heavy atoms of one residue, not the Cα atoms of a chain.

Key triplet generation filters residues against the expected length for each amino acid type, so that only residues meeting the specified condition are processed. The expected lengths are:

Applying these conditions ensures only relevant residues are processed, which supports identification of structural motifs and functional characteristics within protein chains at triplet level.

Publications

Amino Acid TSR was reported in the 2025 study of spike protein and monoclonal antibody interactions:

Tarikul I. Milon, Titli Sarkar, Yixin Chen, Jordan M. Grider, Feng Chen, Jun-Yuan Ji, Seetharama D. Jois, Konstantin G. Kousoulas, Vijay Raghavan and Wu Xu. Development of the TSR-based computational method to investigate spike and monoclonal antibody interactions. Frontiers in Chemistry, 2025, 13, 1395374.

The paper linked at the top of this page, Chen et al. (2025), describes a parallel implementation of TSR comparisons for proteins and amino acids.

Run it with BioTSR

BioTSR's amino_acid_atomic method computes TSR keys from the heavy atoms of one amino acid at a time, each atom labelled by its element. To compute them for Leu15 of ubiquitin:

python -m pip install "biotsr==1.0.0"
biotsr fetch 1UBQ --pdb-dir structures
biotsr run --method amino_acid_atomic --pdb structures/1UBQ.pdb --chain A --residue 15 --residue-name LEU --output results

Leucine's 8 heavy atoms form 56 triangles, with 46 distinct keys, written under the prefix results/1UBQ_A_amino_acid_atomic_LEU_15.

• One residue, one type, or the whole chain: give --residue with --residue-name for one residue, --residue-name alone for one calculation per residue of that name, or neither for one calculation per residue of the chain. Each residue gets its own files, and a sweep has one completion record.
• The expected atom counts: BioTSR keeps this method's rule that a residue is computed only when its heavy-atom count matches the table above. A residue that does not match is declined and the reason recorded; a C-terminal residue with its extra OXT oxygen is one example.
• Related methods: aa_ind and aa_ind_mix pool every atom of one or more residue types, aa_ind_specific pools named residues, and amino_acid_atom_selection uses atoms you name.
• Options: mirror images and size filtering. Amino acid grouping and SSE annotation are Cα options and are refused here.
• From Python: biotsr.analyze(..., method="amino_acid_atomic", residue="15", residue_name="LEU").

The user guide, §5 covers every amino-acid method, and tutorial 02 runs one. For many structures, biotsr batch runs a CSV manifest (tutorial 05), and the command-line guide, §12 has example Slurm job scripts for an HPC cluster.

Comparing with the earlier package

BioTSR's expected atom counts are the earlier package's own, transcribed unchanged. BioTSR labels each atom from the element field of its PDB record (columns 77–78) and has no option to reproduce the earlier package's output files; the known limitations, §1 explain where its standard definitions come from.