Skip to content

v1.0.0

Latest

Choose a tag to compare

@github-actions github-actions released this 11 Nov 12:09
· 1 commit to master since this release
17d2bc2

TMol 1.0.0 provides batched molecular scoring, packing, minimization and FastRelax with PyTorch on CPU and CUDA.

Install

python -m pip install tmol==1.0.0          # CPU / Apple Silicon
python -m pip install "tmol[cu12]==1.0.0"  # CUDA 12-enabled PyTorch
python -m pip install "tmol[cu13]==1.0.0"  # CUDA 13-enabled PyTorch

Add the ligand extra for optional ligand preparation. Requires Python 3.12+ and PyTorch 2.11+. Choose CUDA-enabled PyTorch matching the operator extra. Wheels cover Python 3.12–3.14 on Linux x86_64/aarch64 and Apple Silicon; no local compiler or CUDA toolkit is needed.

Changes

  • Migrate to public AtomWorks 3.1 and Biotite 2, including CIF writing, entity annotations and coordinate precision.
  • Preserve authored input chemistry during ligand hydrogen preparation.
  • Fix PDB identity reads with missing element columns.
  • Refresh installation documentation, README and Colab tutorials, with tutorial code and fixtures pinned to this release.

Known limitation

Legacy LCC nucleotide inputs can fail completion with AtomWorks 3.1.0 after CCD naming and leaving-group changes. Upstream fixes are pending; the narrowly scoped regression is tracked as an expected failure. Supply validated component parameters for affected structures.