Summary
Chemical reaction networks (CRNs) are widely used tools in modeling
biochemical systems such as intracellular processes, metabolic networks
or cell signalling pathways. Being able to produce all the biologically
and chemically important qualitative dynamical features, CRNs have
attracted significant attention in the systems biology community. It is
well-known that the reliable inference of CRN models generally
requires thorough identifiability and distinguishability analysis
together with carefully selected prior modeling assumptions. Here we
present a software toolbox CRNreals that supports the
distinguishability analysis of CRN models using recently published
optimization-based procedures.
Main features
- Algorithmically building the so-called canonical CRN mechanism from kinetic polynomial ordinary differential equations.
- Finding dense and sparse realizations containing the
maximal and minimal nonzero reaction rate coefficients. (We note that
dense realizations give a unique super-structure with a fixed complex
set, and that the CRN structure is unique if and only if the structures
of the dense and sparse realizations are identical.)
- Finding reversible and weakly reversible realizations. The
existence of such realizations has a key effect on the boundedness of
solutions and on the robust stability of the system (depending also on
the deficiency).
- Finding detailed balanced and complex balanced realizations
if they exist. These properties have also important stability
implications.
- Computing dynamically equivalent realizations with the
minimal and maximal number of chemical complexes from a previously
defined complex set.
- Finding the so-called `core' and `non-core' reactions of
the CRN. Core reactions are present in any dynamically equivalent CRN
realization and thus they are indispensable components of the system,
while non-core reactions can (at least mathematically) be substituted
by others.