2.5

Feature release that brings speed optimizations, ChimeraX visualizations, and improvements to rescoring.

Main changes:

• supported Java versions: 17 to 23 (Java 11 is no longer supported)
• speed optimizations
  • ~2x speedup on large datasets (due to faster model inference)
  • ~2x speedup when processing a single file (due to faster model loading)
• added ChimeraX visualizations (thanks to @JavierSanchez-Utges and @elainecmeng for help)
  • new -vis_renderers 'pymol,chimerax' parameter to turn individual visualization renderers on/off
• rescoring improvenemts
  • prank rescore command now generates a predictions.csv file and can serve as a drop-in replacement for prank predict in most scenarios
  • new prank fpocket-rescore command to automatically run fpocket and rescore its predictions
  • new experimental rescoring model prank rescore -c rescore_2024

This version contains the same prediction models as version 2.4 .

2.4.2

Maintenance release with numerous small improvements and bug fixes.

Main changes:

• support Java 21
• support BinaryCIF (bcif) input format
• fix bug where some chains with non-standard residues were not parsed correctly and the part after the non-standard residue was ignored
• add new re-scoring model that uses conservation (prank rescore -c rescore_conservation)
• support re-scoring fpocket predictions in cif format
• support re-scoring PUResNet2.0 predictions

This version contains the same prediction models as version 2.4 .

2.4.1

Maintenance release with numerous small improvements and bug fixes.

Main changes:

• added support for Java 20
• Scripts that execute P2Rank (shell script distro/prank and distro/prank.bat) no longer redirect log (stderr stream) to the file distro/log/prank.log. Instead, they write stderr to the console. This was done to avoid P2Rank writing to the installation directory by default, which may be forbidden on some systems. See issue #59. (thanks to @lilleswing)
• allow absolute file paths in dataset files #58
• improved formatting of stats in output files

If you are already using P2Rank see the list of breaking changes.

This version contains the same prediction models as version 2.4 .

2.4

General-purpose release bringing mmCIF support and new prediction models.

Main changes:

• mmCIF (.cif) fromat support
• new model specialized for predictions on AlphaFold, cryo-EM, and NMR structures (use with -c alphafold)
• new default prediction model with better predictive performance (parametrization and features stay the same but the model was re-trained to take advantage of bugfixes in P2Rank and libraries)
• new prediction model conservation_hmm for upcoming web version: https://p2rank.cz/ using a new sequence conservation pipeline
• added support for Java 17
• updated dependencies to mitigate Log4Shell vulnerabilities

If you are already using P2Rank see the list of breaking changes.

2.3.1

Technical release. Contains new prediction model conservation_hmm for upcoming web version: https://p2rank.cz/ using new sequence conservation pipeline.

2.3

Maintenance release that contains numerous small improvements and bug fixes.

Main changes:

• FASTA export (see export-chains-to-fasta)
• report meaningful errors on invalid command line arguments
• improvements for training and evaluating new models (incl. feature filters)

If you are already using P2Rank see the list of breaking changes.

This version contains the same default prediction model as the version described in the JChem article.

2.2

This is a maintenance release that contains numerous small improvements and bug fixes.

Main changes:

• calibrated probability score for pockets is now included in *_predictions.csv
• fixed small predictive performance regression of v2.1
• fixed bug in conservation based model (results in substantial performance improvement for multi-chain proteins)
• added support for Java 15
• better output formatting
• better logging and error messages
• many improvements for training new models including optimized FasterForest2 classifier and new csv file feature

If you are already using P2Rank see the list of breaking changes.

This version contains the same default prediction model as the version described in the JChem article.

2.1

Changelog:

• Produces a *_residues.csv file with scores and pocket assignments for all protein residues.
• Calculates calibrated probabilities (transformed from scores) for pockets and residues.
• Added support for newer versions of Java (8 to 14).
• Numerous bugfixes and performance optimizations.

This version contains the same default prediction model as the version described in the JChem article.

2.0.1

Bugfix release. Changes:

• fix for github bug #7

2.0 (final)

Final 2.0 release.

Contains the same prediction model as the version described in JChem article with many bugfixes and performance optimizations.