|
Foteini
Makrydaki Degree: Ph.D.,
Chemical Engineering
Advisor: Christos Georgakis
Research Project: Data Driven
Stoichiometric & Kinetic Modeling in Complex
Chemical & Biological Reaction Systems
Project Description:
Complex Reaction Systems consist of a
large number of species whose concentration
change over time is affected by several
interactions that are often not
well-understood due to our limited knowledge
of the stoichiometry and related reaction
kinetics. The reactions involved in the
production of active pharmaceutical
ingredients are one example of a Complex
Reaction System of particular interest in
our research group. If the reactions
related to the production of a drug in the
development stage are studied and understood
reasonable well then the design of the
related manufacturing processes can be done
with less uncertainty and within a shorter
time period.
The methodology under development aims to
achieve the detailed understanding of their
Stoichiometric and Kinetic characteristics
from high throughput experimental data and
the rapid development of an accurate process
model. This model and its accuracy
description can then be used to define the
operational space of the process critical
for FDA approvals and the proper operation
the process.
Our group has developed in the past
systematic approaches in discovering the
stoichiometric models that are consistent
with the experimental data at hand. The
mathematical methods used to acquire the
Stoichiometric Model are Singular Value
Decomposition (SVD) applied to the changes
in the reactants and Structured Target
Factor Analysis (STFA) to test for the
possible reactions taking place. This
methodology is currently tested and extended
to apply in cellular systems, to discover
the active pathways of a metabolic network
that describe interactions between different
proteins whose concentration changes we
measure. At the same time a lot of work has
been done on the development of efficient
nonlinear parameter estimation algorithms if
the kinetic expressions of the involved
reactions are known.
The main objective of this research
project is to develop a general methodology
for the identification of the most
appropriate kinetic form that accurately
describes the experimental data of a complex
reaction. Initially simple kinetic models
are postulated and fitted to the available
data. Methods are under development to
identify the structural model-data mismatch
and thus iterate the kinetic forms to an
incrementally more complex one, until the
structural model-data mismatch is
minimized. If more than one model is
describing the data equally accurately, then
a new set of experiments is designed for
resolving which the more accurate model is.
A final goal of this research investigation
is to be able to do the modeling
concurrently with the data collection;
something that we have termed as online
modeling.
Education &
Experience:
M.S. in Chemical Engineering, National
Technical University of Athens (NTUA), 2004
Research Assistant, Bayer AG - Bayer
Technology Services, Leverkusen, Germany
(2004)
Engineering Trainee, SC Johnson,
Athens, Greece (2003)
Engineering Trainee, Fraunhofer
Institute, Freising, Germany
Funding:
Tufts University |
