dock 0.1.0

Package for Docking.

spock — Standardize, Prepare, Dock

spock is a Python toolkit for structure-based drug design. It bundles a unified molecule/protein storage layer, ligand standardization and preparation pipelines, several docking engines, interaction-fingerprint descriptors, a graph/GNN stack, and an interactive PyMOL GUI — so a target can go from a raw sequence to docked, scored and visualised poses in one place.

Status: active development / beta.

---

Features

Area	What's included
Storage	Unified SpockStorage for ligands, proteins and docked poses, built on the QSPRpred storage interfaces. Per-target stores, compressed pose serialization.
Docking engines	AutoDock Vina (CPU) — multi-core parallel; Vina (CPU, local) — lightweight single-process; Vina-GPU — via the andriusbern/vina-gpu Docker image (NVIDIA runtime). Vina scoring function.
Standardization	Papyrus standardizer and a permissive NaiveStandardizer, with InChI-based identifiers.
Ligand preparation	Dimorphite-DL protonation/tautomer enumeration; Schrödinger LigPrep wrapper.
Descriptors	PLIP protein–ligand interaction fingerprints (PLIPIFP) as a QSPRpred DescriptorSet.
Graphs / GNN	Molecular & complex graph featurizers; PyTorch-Geometric dataset bridge and models ([gnn] extra).
De novo generation	Target-conditioned ligand generation via PCMol from inside the GUI.
Data sources	One-click download of bioactivity data and targets from Papyrus; PDB structure fetching.
GUI	PyMOL-embedded interface for browsing targets, fetching structures, managing ligands, docking, and inspecting poses & interactions (see below).
Parallelism	pebble-based parallel docking, preparation and dataframe pipelines.

---

Installation

Requires Python ≥ 3.10. PyMOL (open-source or incentive build) and AutoDock Vina are expected from the environment — the recommended route is conda/micromamba:

# 1. create an environment with the native deps that don't ship cleanly on PyPI
micromamba create -y -n spock -c conda-forge \
    python=3.11 pymol-open-source pyqt pyqtgraph rdkit openbabel plip

# 2. install spock (editable)
git clone https://github.com/CDDLeiden/spock.git
cd spock
micromamba run -n spock pip install -e .

Optional extras:

pip install -e ".[vina]"    # AutoDock Vina Python bindings (CPU docking)
pip install -e ".[gnn]"     # torch, torch_geometric, wandb, ...
pip install -e ".[pymol]"   # GUI helper stack (pypdb, meeko, docker, plip, ...)
pip install -e ".[dev]"     # linting / tests (pre-commit, ruff, pytest)

For Vina-GPU, an NVIDIA Docker runtime is required; see setup_nvidia_docker.sh.

---

The GUI

Launch the GUI (it runs on PyMOL's own Qt thread) with the installed console script:

spock

or load it as a PyMOL plugin:

pymol spock/gui/addon.py

The interface is organised around a selected target:

Panel	Capabilities
Targets	Browse and fuzzy-search protein targets; download bioactivity data from Papyrus.
Structures	Fetch and load PDB structures for the active target.
Ligands	Add / delete / save / load ligands, download from Papyrus, hover cards with computed properties (incl. SA score).
Docking	Pick a binding-site box ("Get coordinates") and run docking; live progress.
Poses	Per-pose bar with scores, crosshair score plots, load stored poses into the viewer.
Interactions	PLIP protein–ligand contacts rendered directly in PyMOL with a contacts table.
Generator	Load a PCMol model and generate target-conditioned ligands.

All results (structures, ligands, poses, scores) are persisted to the per-target SpockStorage, so closing and reopening a target restores its state.

---

Tutorials

The tutorial/ folder contains notebooks covering storage creation, standardization, ligand preparation and docking benchmarks.

License & citation

See pyproject.toml for authors and project metadata. Part of the CDD Leiden software stack alongside QSPRpred.