A gene regulatory network (GRN) consists of genes, transcription factors, and the regulatory connections between them that govern the level of expression of mRNA and proteins from those genes. The dynamics of a GRN is how gene expression in the network changes over time. GRNmap (Gene Regulatory Network modeling and parameter estimation) uses ordinary differential equations to model the dynamics of small- to medium-scale GRNs. Using a penalized least squares approach, GRNmap estimates production rates, expression thresholds, and regulatory weights for each transcription factor in the network based on timecourse gene expression data (typically generated by DNA microarrays, but amenable to data from other technologies), and then performs a forward simulation of the dynamics of the network. Our approach is described in the paper, Dahlquist, K.D., Fitzpatrick, B.G., Camacho, E.T., Entzminger, S.D., and Wanner, N.C. (2015) Parameter Estimation for Gene Regulatory Networks from Microarray Data: Cold Shock Response in Saccharomyces cerevisiae. Bulletin of Mathematical Biology, 77(8), 1457-1492, DOI: 10.1007/s11538-015-0092-6. Although a GRN from budding yeast was modeled in that paper, GRNmap can be used with any species. Besides what is described in that paper, GRNmap now has an option to use a Michaelis-Menten production function as well as the sigmoidal production function, the ability to input replicate expression data instead of the means for each timepoint, and the option to include data for experiments in which a transcription factor was deleted from the network, among others.

GRNmap is a project of the Loyola Marymount University Biomathematics Group, headed by Dr. Kam Dahlquist, and Dr. Ben G. Fitzpatrick, in collaboration with Dr. John David N. Dionisio. The forerunner to GRNmap was initiated through a collaboration between Dr. Dahlquist and Dr. Erika T. Camacho, who co-mentored undergraduate Nathan C. Wanner (Applied Mathematics '07). Dr. Ben G. Fitzpatrick joined the project in 2007 as part of the NSF-funded UBM (Interdisciplinary Training for Undergraduates in Biological and Mathematical Sciences; 0634613) Project where he and Dr. Dahlquist co-mentored Stephanie D. (Kuelbs) Entzminger (Applied Mathematics '09). The project took its current form with an NSF-funded RUI award (0921038) to Drs. Dahlquist and Fitzpatrick, enabling work by students Alondra J. Vega (Biomathematics '12), Nicholas A. Rohacz (Biochemistry '13), and Katrina Sherbina (Biomathematics '14), and was christened GRNmap in September 2014 with the work of Juan S. Carrillo (Computer Science, Applied Mathematics '16). Subsequent students joined either the coding or data analysis subgroups. The current coding team includes Trixie Anne M. Roque (Computer Science '17) and Chukwuemeka (Eddie) Azinge (Computer Science '19). The current data analysis team includes Natalie E. Williams (Biology '17), Kristen M. Horstmann (Biomathematics '17), Brandon J. Klein (Biology '18) and Magaret J. O’Neil (Biology '18), with recent contributions by Tessa A. Morris (Biomathematics '16), K. Grace Johnson (Biochemistry '17), and Kayla C. Jackson (Spelman College Mathematics '17). GRNmap is featured as a project in the course Biology 398: Biomathematical Modeling/Mathematics 388: Survey of Biomathematics taught by Drs. Dahlquist and Fitzpatrick (see the Spring 2011, Spring 2013, and Spring 2015 course web sites).

  • Running GRNmap from code:
    • The GRNmap code was developed and tested with MATLAB R2014b; it may not function properly with other versions of MATLAB.
    • GRNmap is only compatible with the Windows operating system because of the function it uses to read and write Microsoft Excel spreadsheets. It has only been tested on Windows 7.
    • We recommend running GRNmap with a minimum of 8.00 GB of RAM and a 2.40 GHz processor. GRNmap may be compatible with slower systems; the amount of RAM and processor speed will affect the speed with which GRNmap completes the estimation.

  • Installing and running the GRNmap stand-alone executable:
    • You must have administrator rights or have the MATLAB Runtime Compiler already installed in your machine to use the stand-alone executable.
    • GRNmap is only compatible with the Windows operating system because of the function it uses to read and write Microsoft Excel spreadsheets. It has only been tested on Windows 7.
    • We recommend running GRNmap with a minimum of 8.00 GB of RAM and a 2.40 GHz processor. GRNmap may be compatible with slower systems; the amount of RAM and processor speed will affect the speed with which GRNmap completes the estimation.
The GRNmap v1.10 release adds a minor check to warn the user in the event that running the model with the same input workbook will overwrite output files from a previous run. This release also includes fixes to the test suite.
GRNmap v1.9 entails changes aimed at paying off our technical debt as we move forward to work on newer features. This includes changes to our test suite, and bug fixes alongside their corresponding tests to various issues.
GRNmap v1.8 includes a major revision of the data structure to support missing data and a cleanup of global variables.
GRNmap v1.4.4 includes the following:
  • Routine bug fixes.
  • Implementation of a testing framework. Note that these changes should not affect the user experience, but has resulted in more robust code.
GRNmap v1.4.2 includes the following new features and bug fixes:
  • The MATLAB runtime library has been included in the executable for this release, which was missing from the v1.4 release.
  • GRNmap now supports up to 12 different strains when creating expression graphs.
  • The bug that occurred where the optimization diagnostic figure was not saved when the optimization had fewer than 100 iterations was fixed.
GRNmap v1.4 includes the following new features and bug fixes:
  • The input file can now be located in a directory that contains space characters in the directory name, or on the Desktop.
  • We have standardized the format of the input and output Excel workbooks, making them more readable. See the wiki pages: How-to-format-the-input-file-for-GRNmap-v1.4-and-above and How-to-interpret-the-output-file-for-GRNmap, respectively for details.
    • optimization_parameters variables previously named "time", "simtime", and "Sigmoid" have been renamed as "expression_timepoints", "simulation_timepoints", and "production_function", respectively.
    • The deleted genes/strains are now determined by GRNmap so the "Deletion" parameter is no longer needed in the optimization_parameter sheet.
    • The mean squared error calculation for each individual gene in each strain is now produced in the optimization_diagnostics sheet instead of the sum of squares of error.
  • The current LSE value for each iteration is now shown in the optimization_diagnostics window.
  • The optimization_diagnostic graph is now saved correctly when the number of iterations of the estimation is fewer than 100.
  • Fixed bug when the estimate_params = 0.
  • GRNmap can now perform multiple successive runs for fine tuning of the optimization parameter, alpha. When L_curve is set to 1 in the optimization_parameters worksheet, GRNmap will complete 16 successive runs with the alpha values of 0.8, 0.5, 0.2, 0.1, 0.08, 0.05, 0.02, 0.01, 0.008, 0.005, 0.002, 0.001, 0.0008, 0.0005, 0.0002, and 0.0001. The user supplied initial guesses are used for the first run; all subsequent runs use the output from the previous run as the initial guesses for that run. By plotting the least squared error against the penalty term (found in the optimization_diagnostics sheet of the output file), the user can create an L-curve by which the appropriate alpha value for future estimation runs can be determined from the "elbow" of the curve.
  • The test suite has been updated to reflect recent changes in the package and now uses proper setUp and tearDown methods so that the tests can run faster.
  • All of the test files have been converted to the .xlsx format.
  • The test files have been updated accordingly (i.e., variable names that are obsolete have been removed) after implementing the above changes.
GRNmap v1.2 includes the following new features and bug fixes:
  • Estimation for the threshold parameter in the sigmoidal model for genes with no input has been fixed.
  • Optimization parameters in the test files have been changed to be compatible with the updated code.
  • Updates to the unit testing framework include the following:
    • Test if input files have been read correctly.
    • Test if necessary output worksheets exist.
    • Test if graphs are outputted.
  • "optimizationDiagnostics.jpg" has been changed to "optimization_diagnostics.jpg".
GRNmap v1.0.10 includes the following new features and bug fixes:
  • An optimization_diagnostics sheet with the LSE, Penalty term, min, and iteration count has been created.
    • Computation of the min LSE and SSE for individual genes have also been added to this sheet.
  • Sigmas sheets have been grouped together instead of interlaced with log2_optimized_expression sheets.
  • Graph names have been changed to reflect the deleted strains.
  • The diagnostics figure is now saved along with the other graphs.
  • A bug in the penalty computation of production rates has been fixed.
  • Two demo files have been added with this release:
    • A small network with 4 genes and 6 edges
    • A medium-sized network with 22 genes and 47 edges
GRNmap v1.0.8 includes the following new features and bug fixes:
  • Input/Output worksheet names have been changed.
  • "concentration_sigmas" input worksheet deleted from test_files folder because they are no longer used.
  • Optimized threshold is now outputted when fix_b = 0.
  • Optimized production rates are now outputted when fix_P = 0.
  • Standard deviations of input expressions are now outputted.
  • Documentation of input/output worksheets described above have been updated in the wiki.
  • We have added 16 new test input sheets for each combination of optimization parameters Sigmoid, estimateParams, makeGraphs, fix_b, and fix_P.
GRNmap v1.0.6 includes the following new features and bug fixes:
  • The input Microsoft Excel workbook is now not required to be in the same directory as the GRNmodel code and can now be located in any folder.
  • A bug was fixed where the last graph generated was not being automatically saved; now all graphs are saved in the same folder as the input workbook.
  • The input test files were updated to comply with changes made to the optimization parameter variable names made in release v1.0.4.
  • Ensured that all temporary information is cleared between consecutive estimation or simulation runs.
  • GRNmap now outputs the log2 expression values generated from the forward simulation of the model as individual worksheets for each strain. The worksheets are named with the convention "_log2_optimized_expression".
  • The test file structure and naming convention has been implemented as documented in the wiki.
GRNmap v1.0.4 includes the following new features and bug fixes:
  • The requirements for the "optimization_parameters" worksheet in the input Microsoft Excel workbook have been modified as follows:
    • The name of the parameter iestimate has been replaced with estimateParams.
    • The name of the parameter igraph has been replaced with makeGraphs.
    • These changes were made to make the parameter names more descriptive.
  • The output workbook now contains all of the worksheets from the input workbook.
  • The output workbook now includes worksheets for all of the parameters that were estimated by GRNmap. These output sheets have names that begin with "out_".
GRNmap v1.0.2 includes the following new features and bug fixes:
  • Some known bugs in v1.0 have been fixed.
  • The code has been updated to be compatible with MATLAB 2014b. Version 1.0 would fail when generating graphs because the way that MATLAB treats figure handles had changed.
  • GRNmap now outputs individual worksheets containing the optimized expression values for each strain.
This is the first release of the GRNmap code and executable.

This work is partially supported by NSF award 0921038 (Kam D. Dahlquist and Ben G. Fitzpatrick), the Loyola Marymount University Summer Undergraduate Research Program 2014 (Juan Carrillo) and 2015 (Trixie Roque), the LMU Honors Summer Research Fellowship 2015 (K. Grace Johnson and Natalie Williams), and the LMU Rains Research Assistant Program 2015 (Tessa Morris).

This site uses Google Analytics to count pageviews and downloads to demonstrate usage of our software to interested parties such as our employers and funding agencies. Information about your use of this site is shared with Google. By using this site, you agree to its use of cookies.

This project has been released with a Contributor Code of Conduct. Participants in this project have agreed to abide by its terms. The full Code of Conduct can be read here.