OpenStructure 2.9
Release Notes
- lDDT-PLI now penalizes added model contacts by default.
- Updated unassigned reasons (model and target) to better reflect changes to ligand scoring made in OST 2.8.0.
- Added CSV output (--output-format csv) and by model (rather than target) ligand output (--by-model-ligand-output) to compare-ligand-structures action.
- Improved logging and output of compare-ligand-structures action.
- Residue SetChemType() is now exposed in Python.
- Allow reading of BIRD compounds in PRDCC format in the compound library. Compounds in PRD formats cannot be read and are rejected cleanly.
- The chemdict_tool executable now exits with an error status if no compound were imported.
- The SDF writer populates the program name and time line (2) as per SDF specification.
- The mmCIF reader now refuses to read in files with multiple data blocks (except in fault tolerant mode), and warns about mmCIF files containing more than one model (atom_site.pdbx_PDB_model_num).
- GDT produces slightly higher scores. It's an optimization problem in the end to find the highest possible fraction of CA atoms that superpose within a specified threshold. OpenStructure GDTTS scores are now on average 0.21 points lower (range [0, 100]) than LGA when computed on all CASP15 TS models.
- Enable DockQ on protein-nucleotide and nucleotide-nucleotide interfaces.
- Enable per-atom lDDT scores.
- Optionally compute ICS/IPS scores on trimmed model. Reason for that is scoring with incomplete target structures. The full ICS/IPS scores may
- The SDF reader is now aware of IOProfiles and can read files with invalid bond types (order) from RDKit in fault tolerant mode.
- Several bug fixes and improvements.
Download
Download the latest version of OpenStructure.
Compilation instructions for this version can be found here.