Migrating from PocketVina 0.1

Upgrade an installation

The pip package is now called pocketvina. Version 0.1 was pocketvina-gpu. Remove the old one before you install the new one:

pip uninstall pocketvina-gpu
pip install pocketvina

If you installed 0.2 over 0.1 already, run pip uninstall pocketvina-gpu and then pip install --force-reinstall --no-deps pocketvina.

Installing PocketVina no longer downloads Boost or compiles the engine. P2Rank is downloaded at the first run, and the P2Rank of an older installation in pocketvina/p2rank_2.5/ is still found. You no longer need to set OCL_ICD_VENDORS or other environment variables.

Update your scripts

  • PocketVinaGPU, PocketVinaConfig, P2RankConfig and PocketConfig keep their names and defaults.

  • The Python default is still output_format="files", the 0.1 output folder. Add output_format="atlas" to get the new results folder and the methods best_scores(), show() and select() (Getting started).

  • Ligands are now prepared from their largest fragment with a fixed seed, seed=42, so a run gives the same poses again.

  • To regenerate published 0.1 results, set ligand_prep="legacy". On the command line it is --ligand-prep legacy.

Import 0.1 results

pocketvina import-legacy converts a 0.1 output folder into a results folder. You can then query, view, check and export old runs like new ones:

pocketvina import-legacy output --results results --smiles ligands.csv

output is the output_dir of the old run, and --smiles its ligand CSV. The old folder is not changed. You can try it with the test data of the repository:

pocketvina import-legacy tests/data/legacy_run --results legacy_results --smiles tests/data/smiles.csv
pocketvina query best --results legacy_results

The import reads these files:

output/
  batch_<n>/<name>_cleaned.pdbqt                               ligand inputs of the engine
  batch_<n>_out/<receptor>-pocket<k>-<name>_cleaned_out.pdbqt   engine outputs
  proteins/structures/<receptor>.pdbqt                         receptors
  proteins/pockets/<receptor>.csv                              P2Rank predictions
  configs/<receptor>_pocketrank_<k>_config_batch_<n>.txt       engine configs with the box sizes, if kept

  • The engine outputs are kept unchanged. pocketvina export --results results --what all --format csv --out tables writes the same combined_results.csv again.

  • Without --smiles, the best poses are not converted to SDF. The viewers then show the PDBQT.

  • Running the import again resumes it.

The same in Python:

from pocketvina.io.importer import import_legacy

import_legacy("output", "results", smiles_file="ligands.csv")