David Quigley
Research Summary   -   CV   -   Publications   -   CARMEN   -   equalizer   -   Prostate Cancer
CARMEN is a point-and-click application that permits plotting gene expression, correlation analysis, and differential expression analysis.

What can it do?

The principle functions of CARMEN are:
  • Plot gene expression data
  • Calculate Correlation between genes
  • Generate Correlation networks and export them to Cytoscape
  • Calculate Differential Expression between two conditions
  • Find annotations for genes, or genes which match a GO annotation
This kind of analysis is often performed by people with expertise in the R programming language or other such tools. CARMEN allows researchers who do not have this expertise to take advantage of correlation network analysis approaches.

What do I Need to Run CARMEN?

CARMEN is currently compiled for OS X (Macintosh) and Windows machines. You will need the CARMEN program as well as data to analyze. Each data set consists of gene expression data, information about the probes that make up that gene expression file, and information about the samples in the experiment.

You can get started with data sets we have prepared for you or easily convert your own data for CARMEN.

CARMEN is open-source software and has been compiled for OS X and Windows. Carmen binary programs do not require any other package to run. Source code is on my GitHub repository.

(requires OSX 10.7 or higher)
To install CARMEN for OSX, uncompress and double-click on the Carmen.app file.
Download CARMEN 1.4 for Windows
To install CARMEN for windows, uncompress the zip file and double-click on carmen.exe.

A complete CARMEN user manual (PDF) explains how to use CARMEN.

Development of CARMEN was funded by a grant to Allan Balmain from the NCI Mouse Models of Human Cancer Consortium. If you use CARMEN analysis in your publication, please cite Quigley et al. Cell Reports 2016 (in press).

Statistical calculations in CARMEN are performed behind the scenes by separate command-line programs such as spear, which calculates correlations, and eqtl, which performs eQTL analysis. These programs are multi-threaded C++ code and are many orders of magnitude faster than the equivalent native R code. Command-line programs can be called in reproducible pipelines. Although I generally write my code in R, I perform correlation permutation testing and eQTL analysis using these programs. CARMEN has been used for all of the manuscripts where I've performed correlation or eQTL analysis.

CARMEN is entirely written in C++. The OSX version was compiled in XCode 5.1.1; the Windows version was compiled using Visual Studio 2005. CARMEN uses the Boost library. CARMEN's GUI is a product of the cross-platform wxWidgets library. CARMEN stands for Correlation and Rule Mining Expression Networks, as an earlier version of CARMEN implemented Association Rule Mining methods.