PocketVina Atlas¶

The PocketVina Atlas is a public database of docking results computed in advance: a library of drugs docked with PocketVina against the whole human proteome (the AlphaFold models of human proteins). pocketvina.Atlas reads these results from Python. You can search them, rank them, look at a protein–drug pair in 3D, save its poses and download a part of the Atlas. You run no docking yourself, so no GPU is needed. The website describes the data and how it was made.

To try it in your browser, open the Colab notebook.

Example¶

from pocketvina import Atlas

atlas = Atlas()                               # the public Atlas, read over the internet
atlas.search("imatinib")                      # matching proteins and drugs, best first
atlas.protein("ABL1")                         # gene, name, AlphaFold model, pockets
atlas.drug("imatinib")                        # the drug's identifiers
atlas.best_scores(protein="ABL1", top=20)     # the 20 best drugs on ABL1
atlas.best_scores(ligand="imatinib", top=20)  # the 20 best proteins of imatinib

# the best pose in every pocket of this protein-drug pair
pair = atlas.select(protein="ABL1", ligand="imatinib", pocket="all")
pair.show()                                   # the 3D page with this pair selected
pair.bust()                                   # PoseBusters checks of these poses
pair.export()                                 # the poses as an SDF file
pair.export_protein()                         # the protein as a PDB file

select and the pair's methods are the same as for your own docking results: see Selecting poses and PoseBusters checks.

The 3D page¶

atlas.show(protein="ABL1") shows the protein's page, as show() does for a run. Top hits lists the drugs that score best with a PoseBusters-valid pose, as on the website. Selected pair also has the drug's identifiers, and the page has the protein's information. In Jupyter, Colab or VS Code the page appears below the cell. In a script it is saved as an HTML file, and the path is printed. path="ABL1.html" chooses the file.

This is the page of ABL1. It is interactive: rotate the protein, pick a drug in Top hits or switch pockets. It was made from the small copy of the Atlas that the tests use, so it lists only a few drugs:

3D views describes the parts of the page.

Identifiers¶

protein= takes any identifier of a protein: its gene ("ABL1"), UniProt accession ("P00519"), entry name ("ABL1_HUMAN"), Entrez, HGNC or Ensembl gene ID, a gene synonym or the protein name.

ligand= takes any identifier of a drug: its name or a synonym ("imatinib"), its PVL ID (the Atlas ID of a docked structure), ChEMBL ID, catalogue number, InChIKey (or its first block) or a SMILES of the same molecule.

Case is ignored. If an identifier belongs to several proteins or drugs, the error lists them; then use the UniProt accession or the PVL ID. atlas.search also finds part of a name.

Command line¶

The pocketvina atlas commands do the same in a shell:

pocketvina atlas search imatinib
pocketvina atlas show ABL1 --ligand imatinib --out ABL1_imatinib.html
pocketvina atlas download ABL1 EGFR --out my_atlas

See the command line reference for every option.

Offline use¶

download copies what the Atlas has on some proteins into a folder. Atlas then reads that folder without internet:

atlas.download(proteins=["ABL1", "EGFR"], out="my_atlas")
atlas = Atlas("my_atlas")                       # works offline
atlas.best_scores(protein="ABL1")

The pose files are large. Before it starts, download says how many MB it copies, then shows a progress bar.
A download that stopped goes on where it stopped when you run it again.
The files are checked against the checksums of the release.
In the folder, search and best_scores(ligand=...) list only its proteins.

To read the folder by default, set POCKETVINA_ATLAS to its full path, for example export POCKETVINA_ATLAS=/path/to/my_atlas in a shell. Atlas() and pocketvina atlas then read it.

Where files are kept¶

Files read over the internet are kept in ~/.cache/pocketvina/atlas/ (or $XDG_CACHE_HOME/pocketvina/atlas/), so the second look is fast. You can delete this folder at any time.

Citing¶

If you use the Atlas, cite it as its citation page says. For PocketVina, see Citing and acknowledgements.