<img src="https://github.com/user-attachments/assets/47e16a25-cd58-41eb-9ccd-b40191758d91" alt="py3plex logo" width="400">

Multilayer networks are complex networks with additional information assigned to nodes or edges (or both). This library includes some of the state-of-the-art algorithms for decomposition, visualization and analysis of such networks.

Key Features:

• SQL-like DSL for intuitive network queries with smart defaults
• Multilayer network visualization and analysis
• Community detection and centrality measures
• Network decomposition and embeddings

from py3plex.core import datasets
from py3plex.dsl import Q

# Load a built-in multilayer biological network (~500 nodes, 4 layers)
network = datasets.fetch_multilayer("human_ppi_gene_disease_drug")

# Find key regulator candidates with integrated community detection and uncertainty quantification
master_regulators = (
    Q.communities(                           # Automated community detection
        mode="pareto",                       # Multi-objective Pareto selection
        uq=True,                             # Uncertainty quantification enabled
        uq_n_samples=30,                     # Robustness via 30 perturbed runs
        uq_method="seed",                    # Vary random seeds for stability
        seed=42,                             # Reproducibility
        write_attrs={                        # Attribute names for community info
            "community_id": "community_id",
            "community_stability": "community_stability",
        },
    )
    .nodes()                                 # Switch to node-level analysis
    .node_type("gene")                       # Filter by node type
    .where(degree__gt=3)                     # Remove peripheral nodes
    .uq(method="perturbation", n_samples=100, ci=0.95, seed=42)  # Quantify confidence
    .compute("betweenness_centrality", "pagerank", "degree_centrality")
    .per_layer()                             # Group by layer
        .top_k(30, "betweenness_centrality__mean")  # Top 30 per layer by mean
    .end_grouping()
    .coverage(mode="at_least", k=2)          # Keep nodes that are hubs in ≥2 layers
    .mutate(                                 # Create derived influence score
        score=lambda row: (
            0.5 * row.get("betweenness_centrality__mean", 0.0) +
            0.3 * row.get("pagerank__mean", 0.0) +
            0.2 * row.get("degree_centrality__mean", 0.0)
        )
    )
    .order_by("score", desc=True)
    .limit(20)                               # Final top 20 candidates
    .explain(neighbors_top=5)                # Enrich: community ID, top 5 partners, layers
    .execute(network)
)

df = master_regulators.to_pandas(expand_uncertainty=True, expand_explanations=True)
print(df[["id", "layer", "community_id",
          "betweenness_centrality__mean", "betweenness_centrality_ci95_low",
          "betweenness_centrality_ci95_high", "score", "top_neighbors"]].head(10))

Example output:

    id  layer  community_id  betweenness_centrality__mean  betweenness_centrality_ci95_low  betweenness_centrality_ci95_high     score  top_neighbors
0  252      0            42                      0.025961                         0.021820                          0.030102  0.015577  [{'id': '91', 'weight': 2.3}, {'id': '419', 'weight': 1.9}]
1   91      0            42                      0.024918                         0.020902                          0.028934  0.014951  [{'id': '252', 'weight': 2.3}, {'id': '103', 'weight': 2.1}]
2  419      0            42                      0.024184                         0.020298                          0.028070  0.014510  [{'id': '252', 'weight': 1.9}, {'id': '91', 'weight': 1.7}]
3  103      0            42                      0.023450                         0.019596                          0.027304  0.014069  [{'id': '91', 'weight': 2.1}, {'id': '252', 'weight': 1.8}]
4  375      0            42                      0.022716                         0.018894                          0.026538  0.013628  [{'id': '91', 'weight': 1.8}, {'id': '252', 'weight': 1.6}]

Key features demonstrated:

• Streamlined API: Community detection integrated directly into the DSL query chain
• AutoCommunity: Multi-objective Pareto-optimal selection across algorithms (Louvain, Leiden, etc.)
• Uncertainty Quantification: Confidence intervals for both community detection and centrality measures
• Cross-layer analysis: .coverage(mode="at_least", k=2) keeps only nodes that are hubs in ≥2 layers
• Interpretability: .explain() enriches results with community IDs, top interaction partners, and layer presence
• Composite scoring: Weighted combination of multiple centrality measures for robust ranking

Getting Started

Installation

We recommend using uv for fast, reliable Python environment management:

# Install uv (if not already installed)
curl -LsSf https://astral.sh/uv/install.sh | sh

# Create and activate virtual environment
uv venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

# Install py3plex
uv pip install py3plex

# Or install from source with development dependencies
uv pip install -e ".[dev]"

Alternatively, use pip:

pip install py3plex

Optional Features

Install additional features as needed:

# MCP server for AI agent integration (requires Python 3.10+)
pip install py3plex[mcp]

# Community detection algorithms
pip install py3plex[algos]

# Advanced visualization
pip install py3plex[viz]

# All optional features
pip install py3plex[mcp,algos,viz]

MCP Integration (AI Agents)

py3plex provides a Model Context Protocol (MCP) server for integration with AI coding assistants:

Requirements: Python 3.10 or higher (due to MCP SDK dependency)

# Install with MCP support (Python 3.10+ required)
pip install py3plex[mcp]

# Start MCP server
py3plex-mcp

Configure Claude Desktop (claude_desktop_config.json):

{
  "mcpServers": {
    "py3plex": {
      "command": "py3plex-mcp"
    }
  }
}

The MCP server exposes:

• 7 tools: Load networks, run queries (with DSL v2 support), detect communities, export results, and more
• 3 resources: Complete documentation, DSL v2 reference, and tool schemas
• DSL v2 support: Modern builder API with type hints (Q.nodes().where(degree__gt=5).compute('pagerank'))
• Security-first: Safe file access, automatic output directory, structured errors

See AGENTS.md for complete MCP documentation including DSL v2 examples.

Resources

• Documentation: https://skblaz.github.io/py3plex/
• Technical Book (PDF): Practical Multilayer Network Analysis with Py3plex - Complete handbook (106 pages)
• Examples: examples/ - 170+ example scripts demonstrating usage

License

Py3plex is released under the MIT License.

Note on licensing: Prior to version 1.0, the project was distributed under the BSD-3-Clause license. Starting with version 1.0, the license was changed to MIT to better align with the broader Python scientific ecosystem and simplify contribution and reuse. Both licenses are permissive and OSI-approved.

Citations

@Article{Skrlj2019,
author={Skrlj, Blaz
and Kralj, Jan
and Lavrac, Nada},
title={Py3plex toolkit for visualization and analysis of multilayer networks},
journal={Applied Network Science},
year={2019},
volume={4},
number={1},
pages={94},
abstract={Complex networks are used as means for representing multimodal, real-life systems. With increasing amounts of data that lead to large multilayer networks consisting of different node and edge types, that can also be subject to temporal change, there is an increasing need for versatile visualization and analysis software. This work presents a lightweight Python library, Py3plex, which focuses on the visualization and analysis of multilayer networks. The library implements a set of simple graphica