infomics / ri

RI is a general purpose algorithm for one-to-one exact subgraph isomorphism problem maintaining topological constraints.

RI

Version: 3.6

RI is a general purpose algorithm for one-to-one exact subgraph isomorphism problem maintaining topological constraints.

---

Description

RI is a general purpose algorithm for one-to-one exact subgraph isomorphism problem maintaining topological constraints. It is both a C++ library and a standalone tool, providing developing API and a command line interface, with no dependencies out of standard GNU C++ library. RI works on Unix and Mac OS X systems with G++ installed, and it can be compiled under Windows using Gygwin. Working graphs may be directed, undirected, multigraphs with optional attributes both on nodes and edges. Customizable features allow user-defined behaviors for attribute comparisons and the algorithm's flow.

RI aims to provide a better search strategy for the common used backtracking approach to the subgraph isomorphism problem. It can be integrated with additional preprocessing steps or it can be used for the verification of candidate structures coming from data mining, data indexing or other filtering techniques. RI is able to find graphs isomorphisms, subgraph isomorphisms and induced subgraph isomorphisms. It is distributed in several versions divide chiefly in two groups respectively for static or dynamically changing attributes. All proposed versions are developed taking into account trade-offs between time and memory requirements. Optional behaviors such as stop at first encountered match, processing of result matches, type of isomorphism and additional features may be enabled thanks to high modularity and library's API.

The RI project also aims to provide a comparison of existing exact subgraph matching algorithms on synthetic and real life graphs. An initial collection of datasets is proposed, it includes synthetic and biological graphs but further types of data regarding other research areas (i.e. engineering, computer vision, etc...) are coming. A list of proposed applications is also given.

Please send us an email to get software sources or datasets (see Contacts).

---

Usage

RI - CLI (Command Line Interface)

Before using the RI command line interface of all distributions, please rebuilt it by make -B All versions of RI take in input the same parameters:

 ./ri ISO_TYPE INPUT_FORMAT target_graph pattern_graph    

ISO_TYPE	specify isomorphism
iso	bijective graphs isomorphism.
ind	induced subgraph isomorphism
mono	monomorphism, the classical subgraph matching

INPUT_FORMAT	specify input file format
gfd	directed graphs with attributes only on nodes.
gfu	undirected graphs with attributes only on nodes.
ged	directed graphs with attributes both on nodes and edges.
geu	undirected graphs with attributes both on nodes and edges.
vfu	Sansone et al. file format for labeled directed graphs with attributes only on nodes.

Default graph file format

The RI project provides two graph file format gfu and gfd, respectively for undirected and directed graphs with attributes only on nodes.

Graphs are stored in text files containing one or more items. The current input format allows the description of undirect graphs with labels on nodes.

#[graph_name]
[number of nodes]
[label_of_first_node]
[label_of_second_node]
...
[number of edges]
[node id] [node id]
[node id] [node id]
...

Node ids are assigned following the order in which they are written in the input file, starting from 0. [graph_name] and labels can not contain blank characters (spaces). Labels are case sensitive.

An example of input file is the following:

#my_graph
4
A
B
C
Br
5
0 1
2 1
2 3
0 3
0 2

Indeed, an example of input file in geu format (undirected graph with labels both on nodes and attibutes) is:

#my_graph
4
A
B
C
Br
5
0 1 a
2 1 n
2 3 m
0 3 k
0 2 a

---

Datasets

A biochemical dataset originally used to evaluate the performance of RI and RI-DS is availiabe at this repository.

A further syntetic dataset used to evaluate RI and RI-DS in 2018 is available at the following repository.

---

License

RI is distributed under the MIT license. This means that it is free for both academic and commercial use. Note however that some third party components in RI require that you reference certain works in scientific publications. You are free to link or use RI inside source code of your own program. If do so, please reference (cite) RI and this website. We appreciate bug fixes and would be happy to collaborate for improvements. License

---

Citation

 Bonnici, V., Giugno, R., Pulvirenti, A., Shasha, D., & Ferro, A. (2013).
 A subgraph isomorphism algorithm and its application to biochemical data. 
 BMC bioinformatics, 14(7), S13.

 Bonnici V, Giugno R. 
 On the variable ordering in subgraph isomorphism algorithms. 
 IEEE/ACM transactions on computational biology and bioinformatics. 2016 Jan 7;14(1):193-203.

---

References

if you have used any of the RI project software, please cite the following paper:

Vincenzo Bonnici, Rosalba Giugno, Alfredo Pulvirenti, Dennis Shasha and Alfredo Ferro. A subgraph isomorphism algorithm and its application to biochemical data. BMC Bioinformatics 2013, 14(Suppl 7):S13 doi:10.1186/1471-2105-14-S7-S13.

ri's Issues

RI library questions

How does RI library compare to other subgraph isomorphism matching algorithms, such as VF2/VF3 (here on Github), in terms of speed and memory usage?

If I wanted to construct a graph by adding nodes and edges in code, rather than loading from file, do you provide a way to make this process easy? Something that lets you add nodes and edges to a graph, and hides away the code that handles it.

Thanks.

Clarification on Supported Input Formats in RI

Hello,

First of all, I appreciate with your great work.

I've been examining the RI tool and noticed a discrepancy between the input formats listed in the README and those described in the ri3.cpp file's usage function.

In the README, five input formats are described:

• gfd: directed graphs with attributes only on nodes.
• gfu: undirected graphs with attributes only on nodes.
• ged: directed graphs with attributes on nodes and edges.
• geu: undirected graphs with attributes on nodes and edges.
• vfu: Sansone et al. file format for labeled directed graphs with node attributes only.

However, the usage function in ri3.cpp lists the following formats:

• gfu: undirected graphs with labels on nodes.
• gfd: directed graphs with labels on nodes.
• gfd: directed graphs with a single label on nodes.
• geu: undirected graphs with labels on nodes and edges.
• ged: directed graphs with labels on nodes.

This discrepancy is quite puzzling. Could you please clarify which is the correct set of supported formats? Also, in cases where node labels are entirely absent, can we consider it equivalent to the "direct graphs with one single label on nodes" format as described in the third line of ri3.cpp's usage function?

Thank you for your assistance.

Sincerely,
Syz

Question about bijective graphs isomorphism

Hi, thanks for making your code available.
I don't know much about graph algorithms, so it may be a naive question.

I'm looking for a C++ library that would allow me to iterate over all automorphisms of a graph. (background: I want to superpose two 3D models of small molecules. The graphs of a molecule can have symmetries, so I can't be sure which atoms in the two molecules correspond to each other. Therefore, I want to try all automorphisms of one of the molecules)

I read that this library can find graph isomorphisms. Am I right that it can be used to find automorphisms?
Your 2013 paper compares performance of the subgraph isomorphism algorithms. Do you know how the performance of graph isomorphism in RI compares to other algorithms/libraries such as nauty, saucy or bliss?