Sicifus is a Python library for large-scale structural biology analysis. It ingests thousands of protein structures (CIF/PDB), stores them in a fast Parquet database, and enables comparative analysis, structural alignment, phylogenetics, and physics-based mutation analysis — all on datasets larger than RAM.
Problem: Analyzing thousands of protein structures is slow and memory-intensive. Traditional tools load everything into RAM, limiting dataset size and slowing down queries.
Solution: Sicifus compiles structures into a partitioned Parquet database with lazy evaluation via Polars. This enables:
- Out-of-core processing — Analyze datasets larger than your RAM
- Fast queries — Instant filtering, aggregation, and joins on structural data
- Structural alignment — Sequence-independent alignment using 3Di structural alphabet
- Mutation analysis — ddG prediction, alanine scanning, interface mutations (OpenMM)
- Phylogenetics — RMSD-based trees and clustering
- Ligand analysis — Binding pockets, protein-ligand interactions
Use cases: Comparative structural biology, protein evolution studies, drug discovery, mutation effect prediction, structural genomics.
pip install sicifusFor 2D ligand visualization features (requires RDKit):
pip install "sicifus[viz]"Compile your raw structure files into the database format once.
from sicifus import Sicifus
db = Sicifus(db_path="./my_db")
db.ingest("./data/cif_files", batch_size=100)Align an entire dataset to a reference structure.
# Align all structures to "1atp"
results = db.align_all(reference_id="1atp")
print(results.sort("rmsd").head(5))Generate a structural tree and identify clusters based on RMSD.
# Generate tree and export to Newick
db.generate_tree(output_file="tree.png", newick_file="tree.nwk")
# Annotate clusters with a 2.0 Å RMSD threshold
db.annotate_clusters(distance_threshold=2.0)
print(db.cluster_summary())Analyze binding pockets and interactions.
# Get residue counts in the binding pocket for all structures
pockets = db.get_binding_pockets("LIG", distance_cutoff=8.0)
# Find structures with Tryptophan in the binding pocket
import polars as pl
trp_binders = pockets.filter(pl.col("TRP") > 0)Predict the effect of point mutations on protein stability using open-source tools.
from sicifus import MutationEngine
engine = MutationEngine()
# Calculate total stability
result = engine.calculate_stability("my_protein.pdb")
print(f"Total energy: {result.total_energy:.1f} kcal/mol")
# Mutate Phe-13 to Ala and get ddG
result = engine.mutate("my_protein.pdb", ["F13A"])
print(f"ddG: {result.ddg['F13A']:+.2f} kcal/mol")
# Batch mutations from a CSV
mutations_df = engine.load_mutations("mutations.csv")
results = engine.mutate_batch("my_protein.pdb", mutations_df)
print(results)Full documentation is available at arikat.github.io/sicifus.
MIT