FANMOD+

FANMOD+ is a tool that is based on the original FANMOD (Fast Network Motif Detection) tool - a software tool used for identifying and analyzing network motifs in complex networks. Network motifs are small, recurring patterns or subgraphs that occur more frequently in real-world networks than would be expected in random networks. They are important for understanding the structural and functional building blocks of complex systems, such as biological, social, or technological networks. (Sebastian Wernicke, Florian Rasche, FANMOD: a tool for fast network motif detection, Bioinformatics, Volume 22, Issue 9, May 2006, Pages 1152–1153, https://doi.org/10.1093/bioinformatics/btl038)

FANMOD+ is similar (fork from the fanmod repository) in functionality to FANMOD but extends to support: (1) command line execution with cross-platform build process. (2) more node colors up to 128 and larger motifs - supporting the analysis of modern spatial multiplex single cell omics data. (3) output in LEDA format to support loading the subgraphs in python using networkx lib (https://www.algorithmic-solutions.info/leda_guide/graphs/leda_native_graph_fileformat.html)

Why not use the original FANMOD tool

If you need to extract motifs in a graph with a high number of node colors (more than 15), the original FANMOD tool implementation will reach its limit.

The core change

FANMOD+ increases the space (i.e., number of bits) for the canonical representation from 64-bit fixed allocation to 128-bit dynamic allocation (i.e., physical memory allocation extended when used) (Schäling, 2011).

The FANMODPlus output, input and options are covering only the use case of CISM method (Zamir et al. 2025).

Changes

1. Removing GUI wx dependencies
2. Adding CMAKE compilation files
3. Changed the in-memory data structure that stores the graph canonical labeling of a single motif/graph from 64-bit fixed allocation to 512-bit dynamic allocation (i.e., physical memory allocation extended when used, Schäling B. The boost C++ libraries, 2nd ed. Laguna Hills: XML Press; 2014.).

Build & Run

Prerequisites

Before building FANMOD+, ensure you have the following installed:

• A C++ compiler (e.g., g++) with C++17 support
• CMake (version 3.22 or later)
• Boost (version 1.81.0 or later)
• Git (for cloning the repository)

Linux

mkdir build && cd build
cmake ..
make

For more detailed instructions, especially for building with a local Boost library, please see the Linux Compilation Guide

Linux using Docker

docker build -t fanmod_plus_build/fanmod_plus_build:0.1 .
docker run -it --rm --name=fanmod_plus --mount type=bind,source=${PWD},target="/src" fanmod_plus_build/fanmod_plus_build:0.1 bash

inside the container -

    cd src
    mkdir build && cd build
    cmake ..
    make 

The binary should be created under the build folder (LocalFANMOD)

Note: The minimum required version of glibc is 2.27. You can check your glibc version with ldd --version.

Windows

Compile with visual studio https://learn.microsoft.com/en-us/cpp/build/cmake-projects-in-visual-studio?view=msvc-170

Flags

    -i <path to input graph file>

    Format of the input file should be:
    <left node id> <right node id> <left node color> <right node color>

    -o <path to output file>

    -r <number of null graphs iterations>

    -s <size of motif>

    For instance, 3 is 3-nodes.

    -colored_vertcies <colored vertices>

Output file format

The initial lines of the file contains statistical information

For each motif the output contains information on each candidate subgraph:

ID,Frequency,Count,Mean-Freq,Standard-Dev,Z-Score,p-Value

ID: the canonical unique representation of the subgraph
Frequency: the number of subgraph instances / total number of subgraphs
Count: the number of subgraph instances
Mean-Freq: the mean frequencies of the null graphs
Standard-Dev: the std frequencies of the null graphs
Z-Score: (Frequency-Mean-Freq)/Standard-Dev
p-value: how many times the number of instances is higher than in the null model, divided by the the number of iterations.

-------------------------------------
adjacency matrix with color encoding (not encoded well with more than 9 colors)

leda encoding of the subgraph https://www.algorithmic-solutions.info/leda_guide/graphs/leda_native_graph_fileformat.html
#+leda format
#-leda format

Credit(s)

• The original algorithm was developed by S. Wernicke and F. Rasche

  • See doc/resources/*.pdf for more information.
  • The original source of that implementation was: http://theinf1.informatik.uni-jena.de/motifs/ (no longer available)

• Command line output modifications by Sebastian Bücker

  • See this link;