Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.Sign up
MERLIN is a regulatory network inference algorithm that combines per-gene and per-module regulatory network inference methods. It infers both module memberships of genes, and regulators for individual genes and modules. It is based on a probabilistic graphical model.
Text file with the expression data using the DREAM Challenge format (see File Formats)
l: Text file with the list of regulators (see File Formats). Default: all genes are potential regulators.
c: Text file that specifies the initial module assignments. Default: performs a random partitioning of genes into max[squareroot(n/2),30] clusters, where n is the number of genes in the data. That is the default option will have no more than 30 initial clusters.
h: hierarchical clustering threshold (default 0.6)
p: parameter for sparsity (default -5)
r: parameter for module prior (default 4)
k: max number of regulators that a gene can have (default 300)
o[required] : name of output directory that must exist before running the program
v[required] : number of folds for cross validation (default 1)
You don't need to set parameters that have default values or you can change the default value by setting different values. It is mandatory to set a value for options that don't have default (-data, -reg, -o).
Additional information: Runtime & memory limits## Command-line usage
This section describes how the module can be run locally as a standalone command-line tool (no need to install GenePattern).
Merlin is a C++ tool, which can be downloaded from the repository and compiled on a linux machine.
./merlin -d example/net1_expression.txt -c example/clusterassign.txt -o ./ -l example/net1_transcription_factors.tsv -v 1 -h 0.6 -k 300 -p 5 -r 4
Options are described above.
Source code added to the repository.
To compile the code, type
make in the source directory.
Please cite the following paper when using this module for your work:
- Integrated Module and Gene-Specific Regulatory Inference Implicates Upstream Signaling Networks. Sushmita Roy, Stephen Lagree, Zhonggang Hou, James A. Thomson, Ron Stewart, Audrey P. Gasch.
Sushmita Roy <firstname.lastname@example.org>
Department of Biostatistics and Medical Informatics, Wisconsin Institute for Discovery University of Wisconsin, Madison, WI 53715, USA