1. Parameter ranges file (.prs extension) This file contains the
ranges of different kinetic parameters.
2. Parameters file (_parameter.dat extension) This file contains
the kinetic parameters for each model in the ensemble along
with the number of steady states obtained for that model.
3. Solutions files These files contains the gene expression levels
in each of the steady states obtained for different models.
Steady state expression levels for models exhibiting different
numbers of steady states are stored in different files. For models
with only steady state, the file extension is “_solution_1.dat.”
For models with three steady states, the file extension is “_solution_3.dat.” All gene expression values reported in these files
are log2 normalized.
Descriptions of other output files can be obtained from the
“README.md” file available online with the code.
To determine if epithelial-mesenchymal heterogeneity can
emerge from cell-to-cell variation in kinetic parameters as simulated
using the RACIPE framework, we used a 26-node circuit (Fig. 3;
top panel) which was constructed using Ingenuity Pathway Analysis
(IPA; QIAGEN Inc.) and literature search [62]. The circuit consists
of 17 protein-coding genes and 9 micro-RNAs. The set of proteincoding genes includes transcription factors such as SNAI1, ZEB1,
and TWIST1 whose role as master regulators of EMT is well
characterized [8]. The set also includes EMT-associated biomarkers
such as CDH1 and VIM along with “phenotypic stability factors”
[34] such as GRHL2, OVOL2, and ΔNP63α. The collection of
steady states that can be exhibited by models with the topology of
this EMT circuit was obtained using RACIPE and analyzed using
hierarchical clustering (Fig. 3; bottom panel). As mentioned previously, this collection of steady states is representative of the gene
expression profile of cells in a tumor. The steady states can be
broadly classified into four groups on the basis of expression levels
of the 26 proteins and micro-RNAs in the EMT circuit. Group
1 exhibits high levels of expression of epithelial phenotypeassociated genes including CDH1 along with high levels expression
of EMT inhibitors such as GRHL2 and miR-200. This group thus
represents cells that exhibit an epithelial phenotype. In group
4, EMT drivers such as SNAI1 and ZEB1 are highly expressed
along with high expression of the mesenchymal marker VIM. This
group represents cells that exhibit a mesenchymal phenotype.
Groups 2 and 3 consist of steady states with co-expression of both
epithelial and mesenchymal-associated factors. The expression of
epithelial factors in these groups is lower than the expression of
these factors in the epithelial group (group 1) and the expression of
mesenchymal factors is lower than that in the mesenchymal group
396
Shubham Tripathi et al.
ranges of different kinetic parameters.
2. Parameters file (_parameter.dat extension) This file contains
the kinetic parameters for each model in the ensemble along
with the number of steady states obtained for that model.
3. Solutions files These files contains the gene expression levels
in each of the steady states obtained for different models.
Steady state expression levels for models exhibiting different
numbers of steady states are stored in different files. For models
with only steady state, the file extension is “_solution_1.dat.”
For models with three steady states, the file extension is “_solution_3.dat.” All gene expression values reported in these files
are log2 normalized.
Descriptions of other output files can be obtained from the
“README.md” file available online with the code.
To determine if epithelial-mesenchymal heterogeneity can
emerge from cell-to-cell variation in kinetic parameters as simulated
using the RACIPE framework, we used a 26-node circuit (Fig. 3;
top panel) which was constructed using Ingenuity Pathway Analysis
(IPA; QIAGEN Inc.) and literature search [62]. The circuit consists
of 17 protein-coding genes and 9 micro-RNAs. The set of proteincoding genes includes transcription factors such as SNAI1, ZEB1,
and TWIST1 whose role as master regulators of EMT is well
characterized [8]. The set also includes EMT-associated biomarkers
such as CDH1 and VIM along with “phenotypic stability factors”
[34] such as GRHL2, OVOL2, and ΔNP63α. The collection of
steady states that can be exhibited by models with the topology of
this EMT circuit was obtained using RACIPE and analyzed using
hierarchical clustering (Fig. 3; bottom panel). As mentioned previously, this collection of steady states is representative of the gene
expression profile of cells in a tumor. The steady states can be
broadly classified into four groups on the basis of expression levels
of the 26 proteins and micro-RNAs in the EMT circuit. Group
1 exhibits high levels of expression of epithelial phenotypeassociated genes including CDH1 along with high levels expression
of EMT inhibitors such as GRHL2 and miR-200. This group thus
represents cells that exhibit an epithelial phenotype. In group
4, EMT drivers such as SNAI1 and ZEB1 are highly expressed
along with high expression of the mesenchymal marker VIM. This
group represents cells that exhibit a mesenchymal phenotype.
Groups 2 and 3 consist of steady states with co-expression of both
epithelial and mesenchymal-associated factors. The expression of
epithelial factors in these groups is lower than the expression of
these factors in the epithelial group (group 1) and the expression of
mesenchymal factors is lower than that in the mesenchymal group
396
Shubham Tripathi et al.