ARACNe (Algorithm for the Reconstruction of Accurate Cellular Networks) is a novel algorithm, using microarray expression profiles, that is specifically designed to scale up to the complexity of regulatory networks in mammalian cells, yet general enough to address a wider range of network deconvolution problems. This method uses an information theoretic approach to eliminate the vast majority of indirect interactions typically inferred by pairwise analysis.
On synthetic datasets ARACNe achieves extremely low error rates and significantly outperforms established methods, such as Relevance Networks and Bayesian Networks. Application to the deconvolution of genetic networks in human B cells demonstrates ARACNe’s ability to infer validated transcriptional targets of the c-MYC proto-oncogene.
ARACNe is now available as a Cytoscape app that is integrated in the Cyni Toolbox panel. Click here to download the Cytoscape app.
Available formats for ARACNe2 include the following. (Click on title to access software).
Usage Summary Text File
This file is used by the native aracne2 binaries compiled from C++ source to provide ARACNe2 usage summary. Please copy this file to the same directory as the binary (This can be ignored if the entire source distribution is downloaded)
ARACNe for Windows
PE32 executable for MS Windows (console) Intel 80386 32-bit
ARACNe for GNU/Linux
ELF 64-bit LSB executable, AMD x86-64, version 1 (SYSV), for GNU/Linux 2.6.9, dynamically linked (uses shared libs), for GNU/Linux 2.6.9, not stripped
ARACNe for Mac-OSX
Mach-O 64-bit executable x86_64
ARACNe Java Executable
Java executable jarfile
ARACNe C++ Source
C++ source only with Makefiles
ARACNe Java Source
Java source only with ANT build file
ARACNe Java GUI for Cytospace
Java Graphic User Interface (GUI) for loading adjacency matrices and drawing network diagrams using a built-in Cytoscape plugin. Please set a 'JAVA_HOME' environment variable pointing to your JDK and use the launch_aracne scripts in the distribution to start the application
Documentation and Support
See http://www.nature.com/nprot/journal/v1/n2/abs/nprot.2006.106.html or Download PDF and Supplemental Documents for detailed usage instructions.
aracne [OPTIONS] ... or
java -jar ARACNE-java.jar [OPTIONS]
-i <file> Input gene expression profile dataset
-o <file> Output file name (optional) [*]
-j <file> Existing adjacency matrix (.adj) file
-a <fixed_bandwidth|variable_bandwidth|adaptive_partitioning> Algorithm (fixed_bandwidth | variable_bandwidth | adaptive_partitioning), default: adaptive_partitioning
-k <kernel width> Kernel width (accurate method only), default: determined by program
-b <# bins> No. of bins (fast method only), default: 6
-t <threshold> MI threshold, default: 0
P-value for MI threshold (e.g. 1e-7), default: 1 [**] -e <tolerance> DPI tolerance, default: 1 -h <probeId> Hub gene (only MI w/ hub gene will be computed), default: NONE -r <sample number> Use resampling arrays -s <file> A file containing a list of probes for which a subnetwork will be constructed, default: NONE -l <file> A file containing a list of probes annotated as transcription factors in the input dataset, default: NONE [***] -c <+/-probeId %> Conditional network reconstruction, default: NONE [****] [format: "+24 0.35", "-1973_s_at 0.4"] -f <mean> <cv> Gene filter by the mean and coefficient of variance (cv) of the expression values, default: mean=0, cv=0 -H <ARACNE_HOME> To specify where the ARACNe configuration files locates, default: current working directory --help Display this help and exit [*] If no output file is specified by the user, an output will be automatically generated in the same directory as the input file by appending some of the parameter values, such as kernel width, MI threshold, tolerance and so on, at the end of the input file name, and changing the file extension to ".adj". [**] If the "-t" option is supplied, it will enforce the program to use the specified MI threshold, therefore the "-p" option will be ignored. Otherwise, the program will automatically determines the MI threshold given the p-value. The default, p-value=1, will preserve all pairwise MI. [***] This option is ideal for transcriptional network reconstruction. If provided, DPI will not remove any connection of a transcription factor (TF) by connections between two probes not annotated as TFs. This option is often used in conjunction with '-s', which specifies a list of probes that are either the same or a subset of the probes specified by '-l'. [****] Conditional network reconstructs the network given a specified probe being most expressed or least expressed. In the format that follows "-c", "probeId" indicate the probe to be conditioned on; "+" or "-" specify whether the upper or lower tail of the probe's expression should be used as the condition, and "%" is a percentage between (0, 1) specifying the proportion of samples used as the conditioning subset. Example useage:"-c +24 0.35", "-c -1973_s_at 0.4
Margolin AA, Wang K, Lim WK, Kustagi M, Nemenman I, Califano A. Reverse engineering cellular networks. Nat Protoc. 2006;1(2):662-71.
Basso K, Margolin AA, Stolovitzky G, Klein U, Dalla-Favera R, Califano A. Reverse engineering of regulatory networks in human B cells. Nat Genet. 2005 Apr;37(4):382-90.
Margolin AA, Nemenman I, Basso K, Wiggins C, Stolovitzky G, Dalla Favera R, Califano A. ARACNE: An algorithm for the reconstruction of gene regulatory networks in a mammalian cellular context. BMC Bioinformatics. 2006 Mar 20;7 Suppl 1:S7.
Margolin AA, Nemenman I, Wiggins C, Stolovitzky G, Califano A. On the reconstruction of interaction networks with applications to transcriptional regulation. Accepted in NIPS 2005
Wang K, Banerjee N, Margolin AA, Nemenman I, Basso K, Dalla Favera R, Califano A. Conditional network analysis identifies candidate regulator genes in human B cells. Submitted to RECOMB 2005.