View a markdown version of this page

Run virtual screening with AutoDock Vina on Deadline Cloud - Deadline Cloud

Run virtual screening with AutoDock Vina on Deadline Cloud

The virtual_screening_vina job bundle uses AutoDock Vina to screen large compound libraries against a protein target. The bundle splits a compound library into chunks and docks them in parallel across a fleet of workers, then aggregates and ranks the results by binding affinity.

The pipeline runs four steps:

  1. PrepareReceptor — Converts the protein PDB to PDBQT format with Open Babel.

  2. SplitLibrary — Splits the SDF compound library into N chunks.

  3. DockCompounds — Runs N parallel docking tasks (one per chunk). Each task is idempotent and safe for Spot preemption.

  4. ScoreAndRank — Aggregates all chunk results into a ranked CSV of top hits.

The bundle requires a Linux x86_64 service-managed fleet (Spot recommended), a conda queue environment with openbabel from conda-forge, and AutoDock Vina. Install Vina by building the Build an AutoDock Vina conda package for Deadline Cloud conda recipe into your S3 channel, or by installing the binary through a fleet host configuration script.

Set CompoundLibrary=chembl (default) to automatically download and filter drug-like molecules from ChEMBL, or provide your own SDF file.

From the job_bundles directory, submit the job:

deadline bundle submit virtual_screening_vina \
  -p "ReceptorPdb=receptor.pdb" \
  -p "CompoundLibrary=chembl" \
  -p "MaxCompounds=100000" \
  -p "CompoundsPerChunk=100" \
  -p "CenterX=-10.7" \
  -p "CenterY=12.4" \
  -p "CenterZ=68.8"

For details on building the AutoDock Vina conda package, see Build an AutoDock Vina conda package for Deadline Cloud.