BioTSR

Install

BioTSR 1.0.0 is on PyPI. In an existing Python environment:

python -m pip install "biotsr==1.0.0"

Parallel batch processing is part of this standard installation: joblib, the one runtime dependency, is installed with it, so --jobs works without anything extra.

The optional plotting extra adds matplotlib, which only the clustered heat-map figure needs:

python -m pip install "biotsr[plot]==1.0.0"

BioTSR supports CPython 3.9–3.13 on Linux and macOS. Windows is not supported. The commands are pinned to 1.0.0 so that what you install matches the documentation linked on this page.

Getting started sets up a virtual environment and runs a first calculation. The user guide, §1 covers installation from the GitHub tag and offline installation.

What BioTSR computes

BioTSR runs 20 TSR methods in four groups, all through one shared core for geometry, vertex ordering, binning and key encoding. Each method runs under one standard profile: the TSR science as set by the TSR Research Group.

The modifiers described on the method pages are options of these methods, not separate programs:

• amino-acid grouping — the Cα methods;
• SSE annotation (SSE-TSR) — the Cα methods; it annotates triangles and changes no key;
• mirror images — the Cα and atomic methods, not the cross methods;
• size filtering — every method.

Not every TSR research method is part of BioTSR 1.0. The deep neural network and the feature-selection options used in the hsp70/actin comparison, for example, remain separate research code; see Earlier implementations. Each method page says whether BioTSR runs that method, and how. The capability matrix lists exactly what each method supports, and the reproducibility note gives the command for each published method that BioTSR implements.

Command line and Python

Both interfaces drive the same core. Similarity, clustering and plotting are available from Python only in 1.0.

Documentation

All links point to the documentation released with version 1.0.0.

• Tutorials — eight short walk-throughs, from one protein to clustering; the test suite runs every command in them.
• User guide — installation, every method with an example, and the dataset workflow.
• Command-line guide — every command and option, batches, exit codes, and scripts for HPC clusters.
• Python API guide — running methods, loading results and analysis from Python.
• Capability matrix — which method supports which modifier and arguments.
• Reproducibility note — which command computes which published method.
• Known limitations — what 1.0 does not do, and the input rules it enforces.
• Release notes for 1.0.0

For a first look before the documentation, Learn BioTSR has an interactive lesson and a hands-on notebook built on one ubiquitin example.

BioTSR runs every method under one standard definition and has no option to reproduce the output files of the earlier TSR programs. Where its definition differs from an earlier program's, such as dividing angles by π rather than 3.14 or applying a contact cutoff to both sides of a cross, the method's page says how, and the known limitations, §1 give the general rule.