BRAIN Project

The Biologically Relevant Analysis of Interaction Networks (BRAIN) is a set of algorithms for predicting and analyzing protein domain-peptide ligand interactions based on experimentally known binding evidence (e.g. from protein chip or phage display experiments).

BRAIN can be accessed as a Cytoscape plugin which reads peptide binding profiles and generates interactions displayed as a Cytoscape network.


BRAIN consists of a library or API, and of a Cytoscape plugin.

BRAIN Library

The BRAIN library holds the algorithms and methods for interaction prediction. This library is required to build the BRAIN plugin.

BRAIN Plugin

The BRAIN plugin is a way of running various BRAIN algorithms from the Cytoscape user interface. The plugin runs a prediction analysis and presents the results visually as an interactive Cytoscape network.


The BRAIN library and plugin are available below for download as Java archive (JAR) files and Java source tree tarballs.

Latest Release

BRAIN Plugin: Version 1.0.5 alpha (2007 May 23) BR Build: attachment:brainPlugin.jar BR Source: attachment:BRAIN-Plugin-1.0.5-src.tgz

BRAIN Library: Version 1.0.6 (2007 July 26) BR Build: attachment:brainlib-1.1.jar BR Source: attachment:BRAIN-Library-1.2-src.tgz

Dependencies: Additional JARs required by BRAIN - attachment:brainDeps.tar.gz

Installing the Plugin

System Requirements


Running the Plugin

Notes on Using BRAIN

BRAIN takes one or more query proteins and their respective peptide ligands, builds a PWM-based profile for each protein, searches a protein sequence database for similar profiles, and returns a ranked set of proteins that are predicted to interact with the query proteins.

A query protein and its experimentally determined peptide ligands are represented as a [wiki:/PeptideFile peptide file], in which the peptide binders are listed as an ungapped multiple sequence alignment. The plugin can open a single peptide file, or multiple peptide files that are grouped into a [wiki:/PeptideFile#ProjectFiles project file]. From a peptide file, BRAIN generates a position weight matrix (PWM), or protein profile. The profile constitutes the distributions of amino acids (rows) at each of the positions (columns) of the peptide pattern. In a given column of the PWM, each amino acid is assigned a weight based on its frequency of occurrence at the given position. These weights are normalized to 1.0 to emulate probabilities of occurrence.

BRAIN searches a protein sequence database (in FASTA, SwissProt, GenPept, or EMBL format) for the pattern represented by the protein profile. An interaction score (p-value) is calculated for each hit, and all hits exceeding the user-provided score threshold are reported as predicted interactors of the query protein. The output is provided as a Cytoscape network, enabling further graph-based analysis and exploration of node (protein) and edge (interaction) attributes including interaction score, gene name, motif sequence, and associated OMIM or MorbidMap record (if available).

Release Notes

BRAIN Library 1.1 (2007 July 26)

1.0.6 (2007 June 12)

1.0.5 (2007 May 23)

1.0.4 (2007 May 15)

1.0.3 (2007 April 23)

1.0.2 (2007 April 11)

1.0.1 (2007 February 21)

Future Developments

(in order of priority)

  1. Regular expression filtering of profile search results.
  2. Expose auto-score threshold functionality in Profile Search panel (allow the user to specify automatic score threshold setting instead of having to specify a score threshold themselves).
  3. Expose "regex" search features - new Pattern Search panel similar to Profile Search
  4. New feature: consider aa groups in PWM searching
  5. New feature: consider aa groups in correlation matrix
  6. Design an improved scoring scheme (e.g. take into account surface accessibility) and drop down box to select different scores.


If you have any questions or feedback, please email Moyez Dharsee at

Software/BRAIN (last edited 2008-01-14 23:55:22 by MoyezDharsee)

MoinMoin Appliance - Powered by TurnKey Linux