FPLIP - Fast and Full atom Protein-Ligand Interaction Profiler

A comprehensively restructured and enhanced protein-ligand interaction analysis tool based on PLIP, providing a performance-optimized PLIP-compatible mode and a brand new All-Atom detection mode.

中文文档

---

Overview

FPLIP builds upon PLIP's core functionality with the following major improvements:

1. Performance Optimization

• OpenBabel C-object property pre-extraction optimization: Eliminates repeated Python-C boundary crossing by caching OBAtom properties in Python data structures
• Vectorized distance calculation: Uses scipy distance matrices and numpy array operations instead of pairwise loops
• Cython-accelerated PDB parser: Direct byte array manipulation with pre-compiled lookup tables
• Cuda-accelerated interaction detection: Support CuPy or PyTorch backend and parallel GPU computation for MD trajectory

2. All-Atom Module (New)

• Full-atom interaction detection: No distinction between ligand/receptor; detects all interactions in molecular complexes
• Expanded detection scope:
  • Ligand-protein interactions
  • Protein-protein interactions
  • Intra-protein interactions
  • Intra-ligand interactions (intramolecular H-bonds, etc.)
  • Protein-water, ligand-water, water-water interactions
  • Reamin compatible with DNA/RNA detection
• Chemistry-topology-based charge detection: Unified handling of chemical groups in both proteins and ligands instead of residue-type-based detection in PLIP
• Hydrogen bond-based water bridge detection: Detects water-water interactions based on hydrogen bonds
• Smart self-filtering: Protein residues filter internal interactions; ligands retain intramolecular interactions
• Information preservation strategy: H-bonds filtered by salt bridges are marked as hbond_possible instead of being deleted

---

FPLIP vs PLIP

Detection Scope Comparison

Interaction Type	PLIP	FPLIP All-Atom
Ligand-Protein	✓	✓
Intra-Protein	fragement based	residue-residue based
Intra-Ligand	✗	✓

Chemical Detection Improvements

FPLIP All-Atom uses chemistry-topology-based detection, which is more chemically sound than PLIP's residue-type-based detection:

1. Charge Group Detection

   • PLIP: Based on residue names (e.g., arginine_guanidinium)
   • FPLIP: Based on chemical structure (e.g., guanidinium), applicable to both proteins and ligands

2. Smart Self-Filtering

   • Protein residues: Filter self-interactions within the same residue (no biological meaning)
   • Ligand residues: Retain intramolecular interactions (intramolecular H-bonds are important for ligand conformation)

3. Hydrogen Bond Refinement Strategy

   • PLIP: Directly deletes filtered H-bonds (e.g., H-bonds involved in salt bridges, weaker H-bonds from the same donor)
   • FPLIP: Moves filtered H-bonds to hbond_possible, providing a complete interaction landscape

---

Recommended Use Cases

Scenario	Recommended Mode	Description
Standard drug discovery (ligand-protein)	PLIP-compatible	Compatible with literature results
Comprehensive interaction analysis	All-Atom	Detect all interactions in molecular complexes
Protein-protein interaction research	All-Atom	Analyze protein complex interfaces
Water network research	All-Atom	Direct use of water from MD for water bridge detection
MD trajectory analysis	All-Atom + TrajectoryAnalyzer	Track interaction changes over time
Ligand conformation analysis	All-Atom	Detect intramolecular interactions (intramolecular H-bonds)

Importance in CADD

Intra-Protein Interactions:

• Critical for understanding protein stability and allosteric regulation
• Important for analyzing protein-protein interaction interfaces
• Essential for studying conformational changes upon ligand binding

Intra-Ligand Interactions:

• Intramolecular H-bonds significantly affect ligand conformation and binding affinity
• Understanding internal strain and conformational preferences
• Important for ligand design and optimization

---

Installation

pip install fplip

---

Quick Start

PLIP-Compatible Mode (Ligand-Protein Interactions)

from fplip.structure.preparation import PDBComplex

mol = PDBComplex()
mol.load_pdb('structure.pdb')
mol.analyze()
interactions = mol.interaction_sets[ligand_id].all_itypes

All-Atom Mode (Full-Atom Interaction Detection)

from fplip.all_atom import MoleculeComplex, AtomProperties, UnifiedInteractionDetector

# Load molecule
mol = MoleculeComplex()
mol.load_pdb('structure.pdb')

# Detect all interactions
props = AtomProperties(mol.atom_container)
detector = UnifiedInteractionDetector(mol.atom_container, props, mol.residues)
interactions = detector.detect_all(verbose=True)

# Output includes:
# - 'hbond': Confirmed H-bonds
# - 'hbond_possible': Filtered H-bonds (weaker H-bonds from same donor, H-bonds involved in salt bridges)
# - 'hbond_heavy_atom': Heavy-atom H-bonds when no explicit hydrogens
# - 'hydrophobic', 'saltbridge', 'pistacking', 'pication', 'halogen', 'metal'
# - 'water_bridge': Strict H-bond-based water bridges
# - 'water_bridge_possible': PLIP-style distance+angle water bridges

Convenience Function

from fplip.all_atom import _analyze_complex

interactions, mol, props = _analyze_complex('structure.pdb')

MD Trajectory Analysis

FPLIP provides specialized performance optimizations for MD trajectory analysis:

from fplip.all_atom.trajectory_analyzer import TrajectoryAnalyzer

analyzer = TrajectoryAnalyzer(
    tpr_file='topology.tpr',
    xtc_file='trajectory.xtc',
    gro_file='structure.gro'  # Optional, for residue ID alignment
)

# Load and align
analyzer.load_universe()
analyzer.load_molecule(pdb_str, as_string=True)
analyzer.align_with_mda(frame=0)
analyzer.setup_detector()
analyzer.precompute_detector_once()

# Analyze frame by frame
for frame_idx, interactions in analyzer.iterate_frames(start=0, stop=100, step=1):
    # Process interactions for each frame
    pass

Key Features for MD Analysis:

• One-time initialization: Atom properties and detector caches are computed once
• Fast coordinate updates: Uses KD-tree alignment for rapid coordinate transfer from MDAnalysis
• Water filtering: Automatically filters distant water molecules to focus on relevant hydration shells
• High-order water bridges: Detects water networks connecting two residues through multiple water molecules

Command Line Tools

# Analyze PDB file
python -m fplip.all_atom.cli analyze input.pdb -o results.json

# Generate visualizations
python -m fplip.all_atom.cli static results.json --plot-matrix -o matrix.png

# Launch interactive web visualization
python -m fplip.all_atom.cli interactive results.json --port 8080

---

Dependencies

• Python >= 3.8
• OpenBabel >= 3.0.0 with Python bindings
• NumPy, SciPy
• MDAnalysis (optional, for trajectory analysis)
• Cython (optional, for performance optimization)

---

License

GPLv2

---

Citation

If you use FPLIP in your work, please cite the original PLIP publications:

Adasme,M. et al. PLIP 2021: expanding the scope of the protein-ligand interaction profiler to DNA and RNA. Nucl. Acids Res. (05 May 2021), gkab294. doi: 10.1093/nar/gkab294

Salentin,S. et al. PLIP: fully automated protein-ligand interaction profiler. Nucl. Acids Res. (1 July 2015) 43 (W1): W443-W447. doi: 10.1093/nar/gkv315

---

Maintainer

BHM-Bob_G (bhmfly@foxmail.com)

GitHub: https://github.com/BHM-Bob/fplip