Department of Statistics

A z i z     M i t h a n i

Return to Homepage

A stochastic model for the evolution of metabolic networks with neighbour dependence

Aziz Mithani, Gail M. Preston, Jotun Hein

Bioinformatics 2009, doi: 10.1093/bioinformatics/btp262. Abstract

 

 

Supplementary Material

Java Implementation of the Sampler

Example Data Files

 

Usage:

NetworkEvolution.jar <job> <job_parameters>

Jobs:

  • simulate - Simulate Network Evolution
  • path - MCMC for Path Sampling
  • gibbs - Gibbs Sampler for Parameter Estimation

Job Parameters:

Job: simulate

  1. Evolution Model - 1: Independent Edge Model, 2: Neighbour-Dependent Model
  2. Directory containing network and other files (. for current directory)
  3. Reference Network Filename
  4. Start Network Filename
  5. Core Hyperedges Filename ("" for none)
  6. Prohibited Hyperedges Filename ("" for none)
  7. Insertion Rate
  8. Deletion Rate
  9. Number of Simulation Runs
  10. Number of Iterations per Run
  11. Update Interval for Output Display

Job: path

  1. Evolution Model - 1: Independent Edge Model, 2: Neighbour-Dependent Model
  2. Directory containing network and other files (. for current directory)
  3. Reference Network Filename
  4. Start Network Filename
  5. End Network Filename
  6. Core Hyperedges Filename ("" for none)
  7. Prohibited Hyperedges Filename ("" for none)
  8. Insertion Rate
  9. Deletion Rate
  10. Number of Simulation Runs
  11. Number of Iterations per Run
  12. Number of Iterations for Burn-in Period
  13. Update Interval for Output Display

Job: gibbs

  1. 1. Evolution Model - 1: Independent Edge Model, 2: Neighbour-Dependent Model
  2. Directory containing network and other files (. for current directory)
  3. Reference Network Filename
  4. Start Network Filename
  5. End Network Filename
  6. Core Hyperedges Filename ("" for none)
  7. Prohibited Hyperedges Filename ("" for none)
  8. Insertion Rate (-ve value for random starting value)
  9. Deletion Rate (-ve value for random starting value)
  10. Number of Simulation Runs
  11. Number of Iterations per Run
  12. Number of Iterations for Burn-in Period
  13. Update Interval for Output Display
     

Examples:

Assumption: Data files are in the same directory as the .jar file

  • Simulating network evolution with 10000 iterations, update every 100th iteration, 2 runs
    java -jar NetworkEvolution.jar path 2 . 00030_ref.txt 00030_pae.txt core00030.txt prohibited00030.txt 0.04 0.01 2 11000 100
     
  • MCMC for path sampling with 11000 iterations out of which 1000 are burn-in period, update every 10th iteration, single run
    java -jar NetworkEvolution.jar path 2 . 00030_ref.txt 00030_pae.txt 00030_pst.txt core00030.txt prohibited00030.txt 0.04 0.01 1 11000 1000 10
     
  • Gibbs Sampler with 11000 iterations out of which 1000 are burn-in period, update every 10th iteration, single run
    java -jar NetworkEvolution.jar gibbs 2 . 00030_ref.txt 00030_pae.txt 00030_pst.txt core00030.txt prohibited00030.txt 0.04 0.01 1 11000 1000 10
     
  • Gibbs Sampler with 11000 iterations out of which 1000 are burn-in period, update every 10th iteration, random starting rates, 3 runs
    java -jar NetworkEvolution.jar gibbs 2 . 00030_ref.txt 00030_pae.txt 00030_pst.txt core00030.txt prohibited00030.txt -1 -1 3 11000 1000 10