## CellNetOptimizer (CNO)

** Category** Cross-Omics>Pathway Analysis/Gene Regulatory Networks/Tools

** Abstract** CellNetOptimizer (CNO) is a toolbox for creating logic-based models of signal transduction networks, and training them against high-throughput biochemical data.

CNO can be used independently, or together with DataRail, a complementary MATLAB toolbox for managing, transforming and visualizing data.

All functions are accessible either via scripting or graphical user interfaces (GUIs).

CNO can import models from ProMoT and CellNetAnalyzer.

In the near future, automatic population of CNO models from graphs in the Biological Pathway Exchange (BioPAX) format and those stored in databases such as Pathway Commons.

The models generated from CNO can be stored in DataRail as a data array, making it possible to store models alongside the data used for training.

CNO-based workflow (concepts) for model assembly and calibration --

To train network graphs against data the manufacturer’s developed interoperable CNO and DataRail software that performs five (5) essential tasks:

1) Transforming graphs into compressed Boolean logic superstructures that can be used to compute input-output relationships for the overall network while containing the minimum number of non-identifiable elements;

2) Normalizing biochemical data on the states and activities of signaling proteins so that they can be used to train discrete two-state models;

3) Calibrating models to data based on an objective function that balances goodness of fit with model complexity;

4)) Identifying new links Not present in the starting graph that improve fit to data while marginally increasing model size and false-positive rate; and

5) Manipulating calibrated models to enable their comparison to the starting graph.

In addition to the CNO-based workflow for model assembly and calibration the manufacturer’s recommend a series of computational procedures, involving data and network randomization, derivation of Pareto frontiers, and computation of ROC curves that serve as tests of the quality and reliability of the modeling process.

Workflow (detailed) for creating and calibrating Boolean models using CNO software --

Signed directed graphs are imported into CellNetOptimizer (CNO) and data are imported into DataRail.

The experimental design defines which nodes in the graph are designated and which are undesignated.

The graph is then compressed based on three (3) procedures operating on undesignated nodes.

The compressed graph is transformed into a superstructure that represents a superposition of all Boolean models compatible with the graph.

An optimization algorithm then searches the superstructure for those models that minimize the value of the objective function (theta) for a specific value of the size penalty, (alpha);

Typically this calibration procedure is repeated for multiple values of (alpha).

Optimization is terminated when a predetermined criterion is fulfilled; typically the number of times optimization is performed or when a threshold value for (theta) is reached.

Optimization can then be terminated or new routines initiated to add new edges to the optimized model, followed by another round of calibration aimed at decreasing the value of the objective function.

During edge addition, a higher size penalty (ci) is assigned to edges absent from the initial graph to reflect the fact they are Not supported by prior knowledge.

Once a model has been found, different types of analyses can be performed, such as designing new experiments based on model predictions or comparing models between different cell types.

Moreover, a series of evaluation procedures (as stated above...) should be performed that include cross-validation, ROC curve, and comparison to null models.

For additional detailed info see the following papers:

1) For the “core workflow” implemented for Boolean modeling, see:

J. Saez-Rodriguez, L. G. Alexopoulos, J. Epperlein, R. Samaga, D. A. Lauffenburger, S. Klamt and P. K. Sorger. Discrete logic modeling as a means to link protein signaling networks with functional analysis of mammalian signal transduction, Molecular Systems Biology, 5:331, 2009.

Note: CNO is also called/named “CellNOpt”.

2) For the constrained “Fuzzy Logic” implementation, see:

M. K. Morris, J. Saez-Rodriguez, D. Clarke, P. K. Sorger, D. A. Lauffenburger. Training Signaling Pathway Maps to Biochemical Data with Constrained Fuzzy Logic: Quantitative Analysis of Liver Cell Responses to Inflammatory Stimuli, PLoS Comp. Biol., 7(3): e1001099, 2011.

3) For the MATLAB toolbox, see:

M. K. Morris, I. Melas, J. Saez-Rodriguez. Construction of cell type-specific logic models of signaling networks using CellNetOptimizer. To appear in Methods in Molecular Biology: Computational Toxicology, Ed. B. Reisfeld and A. Mayeno, Humana Press.

*System Requirements*

Contact manufacturer.

*Manufacturer*

- Center for Cell Decision Processes
- Boston, MA, USA
- And
- Department of Systems Biology
- Harvard Medical School
- Boston, MA, USA
- And
- Department of Biological Engineering
- Massachusetts Institute of Technology
- Cambridge, MA, USA

** Manufacturer Web Site**
CellNetOptimizer and/or CNO

** Price** Contact manufacturer.

** G6G Abstract Number** 20703

** G6G Manufacturer Number** 104275