Determining an optimal modularity resolution parameter for MolTI-DREAM

Discover the optimal modularity resolution parameter for your multilayer network when using MolTI-DREAM.

Requirements

This script depends on depends on:

• Python (>=2.6, tested on 3.9.9)
• Pandas (tested on 1.2.0)
• Numpy (tested on 1.19.5)

Input

Add one or more networks with a comma separated file, .csv, in the Networks folder, as seen in the Sample Networks Folder

Run

- Run script: bash optimal_mod.sh 
- Optional arguments: 
    -u upper_limit    set maximum modularity value tested                 [default is 0]  
    -l lower_limit    set minimum modularity value tested                 [default is 50]
    -s steps          set steps to cover the defined range                [default is 5]
    -m min_nodes      set minimal num. of nodes to define a community     [default is 6]
    -r randomizations set number of Louvain randomizations [default 5]

Output

The result is presented at the terminal and at Output directory, determining an optimal modularity resolution parameter as well as the corresponding number of communities and average community size, after filtering for disease modules (communities > 6 nodes). Additional results can be found at the output resulting folder.

Sample Output

molti-output-analysis.txt

mod_param num_communities avg_community_size
5 13 16.0
10 13 14.0
15 12 15.0
20 14 12.0
25 13 12.0
30 13 12.0
35 12 12.0
40 11 11.0
45 11 11.0
50 11 10.0

optimal_mod_parameter.txt

Results:
    Optimal modularity parameter: 25
    - Number of communities: 13
    - Average community size: 12.0

References

Didier G, Valdeolivas A, Baudot A. Identifying communities from multiplex biological networks by randomized optimization of modularity. F1000Res. 2018 Jul 10;7:1042. doi: 10.12688/f1000research.15486.2. PMID: 30210790; PMCID: PMC6107982.