ReactionNetwork

Overview

The ReactionNetwork class is a crucial component in NET-Finder that constructs and analyzes reaction networks based on the collected reaction paths. It provides methods for network visualization, path analysis, and identification of key reaction routes.

Key Features

  • Constructs a reaction network from a set of reaction paths

  • Filters reaction paths based on energy criteria

  • Generates network graphs for visualization

  • Provides methods for analyzing network properties and finding important paths

  • Supports export of network data for further analysis

Basic Usage

Here’s a basic example of how to use the ReactionNetwork class:

from net_finder.net import ReactionNetwork
from net_finder.path import ReactionPath

# Assume we have a list of ReactionPath objects
reaction_paths = [...]  # List of ReactionPath objects

# Create ReactionNetwork instance
network = ReactionNetwork(reaction_paths, energy_cutoff=-22.5)

# Generate and save the network graph
network.generate_graph(output_dir='reaction_network_output')

# Analyze the network
unique_species = network.get_unique_species()
lowest_energy_path = network.get_lowest_energy_path(start_species, end_species)

Parameters

  • reaction_paths (list of ReactionPath): List of reaction paths to construct the network.

  • energy_cutoff (float, optional): Maximum energy threshold for including reactions.

  • structure_diff_threshold (float, optional): Threshold for considering structures as similar (default: 0.1).

Methods

Generates and saves the reaction network graph.

Parameters:

  • output_dir (str): Directory to save the generated graph and related files.

Returns a list of unique species (structures) in the network.

Returns all reaction paths in the network.

Finds the lowest energy path between two species.

Parameters:

  • start_species (ase.Atoms): Starting species structure.

  • end_species (ase.Atoms): End species structure.

Returns:

  • List of ReactionPath objects representing the lowest energy path.

Advanced Usage

For more advanced analysis of the reaction network:

# Find all paths between two species
all_paths = network.find_all_paths(start_species, end_species)

# Identify rate-limiting steps
rate_limiting_steps = network.identify_rate_limiting_steps()

# Calculate network properties
connectivity = network.calculate_connectivity()
centrality = network.calculate_centrality()

# Export network data for external analysis
network.export_to_graphml('network.graphml')

Visualization

The generate_graph method creates a visual representation of the reaction network:

network.generate_graph(output_dir='network_visualization')

This generates: - A PNG file of the network graph - Individual structure files for each unique species - A detailed log of the network generation process

You can customize the appearance of the graph:

network.generate_graph(output_dir='custom_network',
                       node_size=1000,
                       edge_width=2,
                       font_size=12)

Integration with Other Modules

The ReactionNetwork class is typically used as the final step in a NET-Finder workflow:

from net_finder.paths import TransitionStateSearch

# Run transition state searches
ts_search = TransitionStateSearch(initial_structures, calc, dimer_params)
ts_search.run()

# Collect reaction paths
reaction_paths = ts_search.collect_reaction_paths()

# Create and analyze the reaction network
network = ReactionNetwork(reaction_paths, energy_cutoff=-22.5)
network.generate_graph()

# Perform further analysis
key_intermediates = network.identify_key_intermediates()
dominant_pathways = network.find_dominant_pathways(start, end)

Notes

  • The energy cutoff parameter can significantly affect the complexity of the resulting network. Adjust it carefully to balance between completeness and manageability.

  • For large networks, graph generation and some analysis methods may be computationally intensive. Consider using sampling or filtering techniques for very large datasets.

  • The structure comparison used to identify unique species is based on RMSD. Adjust the structure_diff_threshold if needed for your specific system.

  • While the generated graph provides a good overview, consider using specialized network analysis tools for more advanced analyses of large or complex networks.