306 B. Cummins et al.
be considered a representation of such unchecked progression through the cell
cycle. While the original network does not exhibit this behavior, the extended
network does. From this perspective,X 7 andX 8 can be viewed as a prediction
of new oncogenes, or product of oncogenes.
4 Code Availability
DSGRN is publicly available [ 22 ]. The website contains complete documentation
including installation instructions, access to the GitHub repository, a graphical
interface that allows user to construct the network to be analyzed, and a collec-
tion of databases for networks that have already been computed. DSGRN can
work in two different modalities.
1.Exploration of the parameter space.In this mode, which is described
in the Documentation directory on the DSGRN website [ 22 ], the program
computes an SQL database from a network file. (See the Documentation on
Network Specification Files for the format of input network files.) Included
with this software is a command line tooldsgrnthat accesses meta-data about
the network directly from the network file without computing the whole data-
base. Many precomputed databases can be viewed in the Databases directory
on the DSGRN website [ 22 ]. The output of each database is presented as a
collection of Morse graphs in descending order based on the number of para-
meter nodes where this Morse graph is observed. Selected filters are available
that allow the user to limit the types of dynamics that are visible. To view
a database that is not pre-computed, go tohttps://dsgrn.comand request
an account. Follow the documentation on that page and also see the tutorial
Bistable Repressilator Jupyter notebook in the top level Tutorial directory
in the DSGRN GitHub repository (the repository is linked from [ 22 ]). These
documents explain how to compute a database on the server and view it on
a personal website.
2.Computation at selected parameter nodes.The installation described
in the Documentation directory on the DSGRN website [ 22 ] downloads, but
does not install, a package of Python tools that can be used to compute the
State Transition Graph and other details about the dynamics at parame-
ter nodes of the parameter graph. The parameters of interest can be chosen
from examining Morse graphs in step 1. The installation instructions for the
Python package and its dependencies are described in the README file
in DSGRN/software/Python/ on GitHub. Furthermore, Jupyter Notebooks
with tutorials on the python tools (DSGRNGettingStarted) and on prepro-
grammed SQL queries wrapped in python (QueryTutorial) can be found in
the GitHub repository in DSGRN/software/Python/doc.There are two main limitations on the size of the network that can be com-
puted. One limitation is the type of node in the network; currently, nodes with
up to 3 inputs and 5 outputs can be computed, with selected higher orders avail-
able. The reason is that the structure of the parameter graph corresponding to