Installation
Sicifus requires Python 3.9+.
Recommended Installation (Conda)
We strongly recommend using a fresh Conda environment to avoid version conflicts (especially with NumPy/Numba).
# 1. Create a fresh environment
conda create -n sicifus_env python=3.11
conda activate sicifus_env
# 2. Install core numerical stack (prevents binary conflicts)
# Note: We pin numpy<2.0 because Numba is not yet compatible with NumPy 2.0+
# Note: We install llvmlite/numba via conda to avoid build errors on some systems
conda install -c conda-forge "numpy<2.0" scipy matplotlib networkx polars gemmi openmm pdbfixer xtb llvmlite numba
# 3. Install Sicifus
pip install ".[all]"
Fast Installation (uv)
If you prefer uv for lightning-fast package management:
# 1. Create venv
uv venv sicifus_env
source sicifus_env/bin/activate
# 2. Install dependencies
# Note: xtb still needs to be installed via conda or available in PATH
uv pip install "numpy<2.0"
uv pip install ".[all]"
Standard Installation
Dependencies
polars: Fast DataFrames.gemmi: Macromolecular structure parsing.numpy&scipy: Numerical computing and alignment algorithms.networkx: Graph/Network analysis.matplotlib: Visualization.rdkit(optional): 2D ligand depiction. Install withpip install sicifus[viz].
Installing with RDKit
For 2D ligand visualization features, you need RDKit.
Installing with Energy & Mutation Tools
The [energy] extra installs openmm and pdbfixer, which power both the xTB energy scoring pipeline and the Mutation & Stability Engine (in silico mutagenesis, ddG, alanine scanning, etc.).
# Python dependencies (covers both xTB scoring and mutation engine)
pip install ".[energy]"
# xTB binary (only needed for ligand energy scoring, not for mutations)
conda install -c conda-forge xtb
The mutation engine works entirely through OpenMM and PDBFixer — no external binaries required.
Full Installation
To install everything: